lang
stringclasses
8 values
year
stringclasses
27 values
title_en
stringlengths
4
253
title_cs
stringlengths
0
251
abstract_en
stringlengths
49
5.12k
abstract_cs
stringlengths
33
3.92k
authors
sequencelengths
1
577
s2_url
stringlengths
79
79
title_fr
stringclasses
3 values
abstract_fr
stringclasses
3 values
title_ru
stringclasses
13 values
abstract_ru
stringclasses
11 values
title_sk
stringclasses
2 values
abstract_sk
stringclasses
2 values
title_de
stringclasses
4 values
abstract_de
stringclasses
4 values
title_dsb
stringclasses
2 values
abstract_dsb
stringclasses
2 values
title_lt
stringclasses
1 value
abstract_lt
stringclasses
1 value
en
2015
Giving a Sense: A Pilot Study in Concept Annotation from Multiple Resources
Získání významu: pilotní studie anotace konceptů na základě více zdrojů
We present a pilot study in web-based annotation of words with senses coming from several knowledge bases and sense inventories. The study is the first step in a planned larger annotation of “grounding” and should allow us to select a subset of these “dictionaries” that seem to cover any given text reasonably well and show an acceptable level of inter-annotator agreement.
V článku překládáme pilotní studii zaměřenou na anotaci významu slov získaných z několika informačních zdrojů. Studie je prvním krokem z zamýšlené anotací "ukotvení".
[ "Roman Sudarikov", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/57daedbeade27cdbdefc8a6f7916265f66b758a0/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Giving a Sense: A Pilot Study in Concept Annotation from Multiple Resources
Získání významu: pilotní studie anotace konceptů na základě více zdrojů
We present a pilot study of a web-based annotation of words with senses. The annotated senses come from several knowledge bases and sense inventories. The study is the first step in a planned larger annotation of grounding and should allow us to select a subset of the sense sources that cover any given text reasonably well and show an acceptable level of inter-annotator agreement.
Představujeme pilotní experimenty s anotací významů slov ve webovém anotačním prostředí. Významy přitom pocházejí z několika různých zdrojů. Experimenty jsou prvním krokem v plánované větší anotaci a měly by nám pomoci s výběrem podmnožiny studovaných zdrojů tak, aby pokrývaly veškerý text v co největším rozsahu a současně vykazovaly přijatelnou úroveň shody mezi anotátory.
[ "Roman Sudarikov", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/57daedbeade27cdbdefc8a6f7916265f66b758a0/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
TeamUFAL: WSD+EL as Document Retrieval
TeamUFAL: WSD+EL jako Document Retrieval
This paper describes our system for SemEval- 2015 Task 13: Multilingual All-Words Sense Disambiguation and Entity Linking. We have participated with our system in the sub-task which aims at monolingual all-words disambiguation and entity linking. Aside from system description, we pay closer attention to the evaluation of system outputs
V članku se popisuje system pro SemEval- 2015 Task 13: Multilingual All-Words Sense Disambiguation and Entity Linking. My jsme se zameřili na monolingvalní disambiguaci a propojeni entit.
[ "Roman Sudarikov", "Petr Fanta", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/6d2bbc2d36f3f5e713c04b32263ba975112b1ec1/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
What a Transfer-Based System Brings to the Combination with PBMT
Přínosy systému založeného na transferu pro PBMT
We present a thorough analysis of a combination of a statistical and a transfer-based system for English->Czech translation, Moses and TectoMT. We describe several techniques for inspecting such a system combination which are based both on automatic and manual evaluation. While TectoMT often produces bad translations, Moses is still able to select the good parts of them. In many cases, TectoMT provides useful novel translations which are otherwise simply unavailable to the statistical component, despite the very large training data. Our analyses confirm the expected behaviour that TectoMT helps with preserving grammatical agreements and valency requirements, but that it also improves a very diverse set of other phenomena. Interestingly, including the outputs of the transfer-based system in the phrase-based search seems to have a positive effect on the search space. Overall, we find that the components of this combination are complementary and the final system produces significantly better translations than either component by itself.
Představujeme podrobnou analýzu kombinace systému založeného na transferu TectoMT se statistickým systémem Moses pro překlad mezi angličtinou a češtinou. Popisujeme možnosti zkoumání této kombinace systémů jak pomocí ruční, tak i automatické evaluace. Přesto, že výstupy TectoMT často obsahují chyby, Moses dokáže z jeho výstupů vybrat vhodné části. V mnoha případech pak TectoMT poskytne nové, užitečné překladové varianty, které jsou pro statistickou komponentu jinak nedosažitelné, navzdory velikosti trénovacích dat. Naše analýzy potvrzují, že TectoMT napomáhá dodržení gramatické shody a požadavků valence, ale zároveň zlepšuje překlad rozmanité škály jazykových jevů. Zahrnutí výstupů systému založeného na transferu do frázového překladu má také zřejmě pozitivní vliv na prohledávaný prostor hypotéz. Zjišťujeme, že jednotlivé komponenty této kombinace jsou komplementární a výsledný systém překládá signifikantně lépe než kterákoliv jednotlivá komponenta.
[ "Aleš Tamchyna", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/124aa8493633100a463adcbb05bdff1ff7f4ae73/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
A Discriminative Model for Semantics-to-String Translation
Diskriminativní model pro překlad ze sémantické reprezentace
We present a feature-rich discriminative model for machine translation which uses an abstract semantic representation on the source side. We include our model as an additional feature in a phrase-based decoder and we show modest gains in BLEU score in an n-best re-ranking experiment.
Představujeme diskriminativní model s bohatou sadou rysů pro strojový překlad, který využívá abstraktní sémantickou reprezentaci na zdrojové straně. Náš model využíváme jako nový rys ve frázovém překladači a dosahujeme mírných zlepšení BLEU skóre v experimentu n-best reranking.
[ "Aleš Tamchyna", "Chris Quirk", "Michel Galley" ]
https://www.semanticscholar.org/paper/8363b98c46e484e213c212626ae11d4f89d95026/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Czech Aspect-Based Sentiment Analysis: A New Dataset and Preliminary Results
Postojová analýza aspektů v češtině: nová datová sada a předběžné výsledky
This work focuses on aspect-based sentiment analysis, a relatively recent task in natural language processing. We present a new dataset for Czech aspect-based sentiment analysis which consists of segments from user reviews of IT products. We also describe our work in progress on the task of aspect term extraction. We believe that this area can be of interest to other workshop participants and that this paper can inspire a fruitful discussion on the topic with researchers from related fields.
V této práci se zaměřujeme na postojovou analýzu aspektů, což je relativně nová úloha z oblasti počítačového zpracování přirozených jazyků (NLP). Představujeme novou datovou sadu v češtině, která sestává z uživatelských hodnocení produktů z oblasti IT. Zároveň popisujeme průběh naší práce na automatické extrakci aspektů. Věříme, že tato oblast může účastníky workshopu zaujmout a že tento příspěvek podnítí diskuzi na toto téma s výzkumníky z příbuzných oborů.
[ "Aleš Tamchyna", "Ondřej Fiala", "Kateřina Veselovská" ]
https://www.semanticscholar.org/paper/f56e63471a1773f1068e3c962b2cbf29d2f28608/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Parsito
Parsito
Parsito is a fast open-source dependency parser written in C++. Parsito is based on greedy transition-based parsing, it has very high accuracy and achieves a throughput of 30K words per second. Parsito can be trained on any input data without feature engineering, because it utilizes artificial neural network classifier. Trained models for all treebanks from Universal Dependencies project are available (37 treebanks as of Dec 2015).
Parsito je rychlý závislostní parser napsaný v C++ vydaný jako open-source. Parsito je založené na transition-based parsingu, má vysokou úspěšnost a dosahuje rychlosti 30 tisíc slov za sekundu. Parsito lze natrénovat na libovolných vstupních datech, bez nutnosti navrhovat jazykově závislé rysy, protože používá klasifikátor založený na neuronových sítích. K dispozici jsou natrénované modely pro všechny treebanky z projektu Universal Dependencies (37 treebanků k prosinci 2015).
[ "Milan Straka" ]
https://www.semanticscholar.org/paper/4cfc22024f843c5a1a6f76b56b1d1c067ea6bfc3/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
MorphoDiTa
MorphoDiTa
MorphoDiTa: Morphological Dictionary and Tagger is an open-source tool for morphological analysis of natural language texts. It performs morphological analysis, morphological generation, tagging and tokenization and is distributed as a standalone tool or a library, along with trained linguistic models. In the Czech language, MorphoDiTa achieves state-of-the-art results with a throughput around 10-200K words per second. MorphoDiTa is a free software under LGPL license and the linguistic models are free for non-commercial use and distributed under CC BY-NC-SA license, although for some models the original data used to create the model may impose additional licensing conditions.
MorphoDiTa (morfologický slovník a tagger) je open-source nástroj pro morfologickou analýzu textů v přirozených jazycích. Provádí morfologickou analýzu, morfologické generování, tagování a tokenizaci a je distribuován jako samostatný nástroj nebo jako knihovna spolu s natrénovanými lingvistickými modely. V českém jazyce dosahuje MorphoDiTa state-of-the-art výsledků s rychlostí 10-200 tisíc slov za sekundu. MorphoDiTa je svobodný software pod LGPL licencí a jazykové modely jsou zdarma pro nekomerční použití a jsou distribuovány pod CC BY-NC-SA licencí, i když u některých modelů mohou původní data použitá k vytvoření modelu implikovat další licenční podmínky.
[ "Milan Straka", "Jana Straková" ]
https://www.semanticscholar.org/paper/442efc550ef93631746f1182836c9d9345f939c0/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
NameTag
NameTag
NameTag is an open-source tool for named entity recognition (NER). NameTag identifies proper names in text and classifies them into predefined categories, such as names of persons, locations, organizations, etc. NameTag is distributed as a standalone tool or a library, along with trained linguistic models. In the Czech language, NameTag achieves state-of-the-art performance (Straková et. al. 2013). NameTag is a free software under LGPL license and the linguistic models are free for non-commercial use and distributed under CC BY-NC-SA license, although for some models the original data used to create the model may impose additional licensing conditions.
NameTag je open-source nástroj pro rozpoznávání jmenných entity (Named Entity Recognition - NER). NameTag identifikuje vlastní jména v textu a zařazuje je do předem definovaných kategorií, jako jsou názvy osob, míst, organizací, atd. NameTag je distribuován jako samostatný nástroj nebo jako knihovna spolu s natrénovanými lingvistickými modely. V českém jazyce dosahuje NameTag state-of-the-art výkonu (Straková et al.,. 2013). NameTag je svobodný software pod LGPL licencí a jazykové modely jsou zdarma pro nekomerční použití a jsou distribuovány pod CC BY-NC-SA licencí, i když u některých modelů mohou původní data použítá k vytvoření modelu implikovat další licenční podmínky.
[ "Milan Straka", "Jana Straková" ]
https://www.semanticscholar.org/paper/d9c6cb8ac25829489c4c77ff3c901a305c343a8c/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Parsing Universal Dependency Treebanks using Neural Networks and Search-Based Oracle
Parsování Universal Dependency korpusů pomocí neuronových sítí a prohledávacího orákula
We describe a transition-based, non-projective dependency parser which uses a neural network classifier for prediction and requires no feature engineering. We propose a new, search-based oracle, which improves parsing accuracy similarly to a dynamic oracle, but is applicable to any transition system, such as the fully non-projective swap system. The parser has excellent parsing speed, compact models, and achieves high accuracy without requiring any additional resources such as raw corpora. We tested it on all 19 treebanks of the Universal Dependencies project. The C++ implementation of the parser is being released as an open-source tool.
V článku popisujeme přechodový neprojektivní závislostní parser používající klasifikátor založený na neuronových sítích, který nevyžaduje tvorbu rysů. Dále představujeme nové přechodové orákulum, které zvyšuje úspěšnost parseru porovnatelně s dynamickým orákulem, ale je použitelné pro každý přechodový systém, jako například neprojektivní systém s operací swap. Parser je velmi rychlý, jeho modely kompaktní, přičemž dosahuje vysoké úspěšnosti bez potřeby dalších zdrojů jako například korpusů s čistým textem. Parser jsme otestovali na všech 19 korpusech z projektu Universal Dependencies. Implementaci parseru v jazyce C++ uvolňujeme jako open-source.
[ "Milan Straka", "Jan Hajič", "Jana Straková", "Jan Hajič, jr." ]
https://www.semanticscholar.org/paper/6e7f07aadc0f1931abfdff3403d543d1879624ce/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Data Citation Box
Citovač dát
Citing submissions is important but citing data submissions has not an established format yet. We have created a citation service based on DSpace OAI-PMH endpoint implementation which returns citations of resources specified by PID and in desired format like simple html styled text. We display the citation box in DSpace item view but also in external applications.
Vytvorili sme citačnú službu na protokole OAI-PMH pre digitálnu knižnicu DSpace.
[ "Jozef Mišutka" ]
https://www.semanticscholar.org/paper/a007fd97885ccce88a435754f1213195afd3d57b/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
How to set up a CLARIN B-centre using a modified DSpace: DSpace & becoming a B centre
Ako postaviť CLARIN-B centrum s použitím DSpace
This talk presents the steps required for becoming a CLARIN-B centre using LINDAT/CLARIN digital library based on DSpace.
CLARIN centrum sa môže nazývať CLARIN-B centrum iba ak spĺňa špecifické kritériá. Digitálny repozitár LINDAT/CLARIN postavený na DSpace spĺňa väčšinu týchto požiadaviek.
[ "Jozef Mišutka", "Amir Kamran" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Attractive and Effective Data Sharing
Atraktívne zdieľanie dát
Our experience show that submitters of data to our repository request specific features in contrast to the established ones. We have identified four such features which we implemented into our repository based on DSpace. Three features are described in this paper in more detail. The last one is described in a separate paper.
Identifikovali sme štyri základné vlastnosti, ktoré spravili zdieľanie dát v našom repozitári atraktívnejšie.
[ "Jozef Mišutka", "Amir Kamran", "Ondřej Košarko" ]
https://www.semanticscholar.org/paper/71dfa736795e1bc367def3e7e05e74572f18f54b/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Sharing and reusing data legally
Legálne zdieľanie dát
The necessity to share and preserve data and software is becoming more and more important. Without the data and the software, research cannot be reproduced and tested by the scientific community. Making data and software simply reusable and legally unequivocal requires choosing a license for data and software which is not a trivial task. We describe a legal/licensing framework which implements the complete support for licensing submissions.
Zdieľanie dát sa stáva neoddeliteľnou súčasťou vedeckej práce. Pripraviť dáta na zdieľanie vyžaduje vybranie správnej licencie. Popisujeme framework, ktorý takéto zdieľanie umožňuje.
[ "Jozef Mišutka", "Amir Kamran", "Ondřej Košarko", "Michal Sedlák", "Pavel Straňák" ]
https://www.semanticscholar.org/paper/3e18b79e8c0a5cf26a8da190cf8a54de2b66530b/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Consistent User Experience and Cross-referencing inside LINDAT/CLARIN infrastructure
Jednotný uživatelský prožitek a odkazování v infrastruktuře LINDAT/CLARIN
With an increasing amount of data and services available in the LINDAT/CLARIN infrastructure comes the need to properly cite, share, cross-reference and gather usage statistics. We introduce a framework that ensures a consistent look and the mentioned functionality.
S přibývajícím množstvím dat a služeb dostupných v infrastruktuře LINDAT/CLARIN roste potřeba správně citovat, sdílet, odkazovat a získávat statistiky využití. Představení frameworku, který zajišťuje jednotný vzhled a zmíněnou funkcionalitu.
[ "Jozef Mišutka", "Ondřej Košarko", "Amir Kamran", "Michal Sedlák" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
CsEnVi Pairwise Parallel Corpora
CsEnVi Pairwise Parallel Corpora
CsEnVi Pairwise Parallel Corpora is a collection of two parallel corpora: an English-Vietnamese one and a Czech-Vietnamese one. The corpora contain translations of movie and TED subtitles that were already available. The corpora were cleaned using our semi-automatic filtering and are provided aligned at the sentence level.
CsEnVi Pairwise Parallel Corpora je soubor dvou paralelních korpusů, anglicko-vietnamského a česko-vietnamského, sestavených z dostupných filmových titulků a titulků přednášek TED. Korpusy byly očištěny naší poloautomatickou filtrací a jsou zarovnány po větách.
[ "Tam Hoang", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/7b675d2f639c0da9e3b882b352bab5d0e39ae5c5/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
TmTriangulate: A Tool for Phrase Table Triangulation
TmTriangulate: Nástroj pro spojování frázových tabulek triangulací
Over the past years, pivoting methods, i.e. machine translation via a third language, gained respectable attention. Various experiments with different approaches and datasets have been carried out but the lack of open-source tools makes it difficult to replicate the results of these experiments. This paper presents a new tool for pivoting for phrase-based statistical machine translation by so called phrase-table triangulation. Besides the tool description, this paper discusses the strong and weak points of various triangulation techniques implemented in the tool.
Představujeme nástroj pro spojování frázových tabulek prostřednictvím prostředního, pivotního, jazyka.
[ "Tam Hoang", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/6221db9b6b968154e2e0dd10edd45de95e1acab2/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
WMT2013 Test Set in Vietnamese
Testset WMT2013 ve vietnamštině
The file contains the standard WMT 2013 test set, manually translated from English to Vietnamese, thus extending the WMT 2013 multi-parallel set of languages (English, Czech, German, French, Spanish, Russian).
Soubor obsahuje standardní test set WMT z roku 2013, ručně přeložený z angličtiny do vietnamštiny. Soubor tak představuje obohacení dosavadní multiparalelní kolekce (anglicky, česky, německy, francouzsky, španělsky a rusky) o nový jazyk.
[ "Tam Hoang", "Ondřej Bojar" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Constraint-based semantic parsing
Sémantická analýza textu založená na omezujících podmínkách
We introduce a model and an algorithm for semantic parsing in λ-calculus formalism which is based on Constrained Conditional Model framework (Roth and Yih, 2005). We also introduce an approach for learning the constraints of our model directly from knowledge graph which scales up for new domains easily. In this ongoing project we optimize our model on a dataset using integer linear programming. Our project includes two parts and our results in the first part show a meaningful improvement over previous models.
Představujeme model a algoritmus pro sémantickou analýzu ve formalismu lambda-kalkulus, který je založen na přístupu Constrined Conditional Model.Dále představujeme metodu pro učení omezujících podmínek přímo ze znalostních grafů. V našem probíhajícím průzkumu optimalizujeme model pomocí lineárního programování.
[ "Ahmad Aghaebrahimian" ]
https://www.semanticscholar.org/paper/b4b3d45a2498e47aae0a52014a5e7974f984e2df/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Machine Learning for Semantic Parsing in Review
Strojové učení pro sémantický parsing – přehled
Spoken Language Understanding (SLU) and more specifically, semantic parsing is an indispensable task in each speech-enabled application. In this survey, we review the current research on SLU and semantic parsing with emphasis on machine learning techniques used for these tasks. Observing the current trends in semantic parsing, we conclude our discussion by suggesting some of the most promising future research trends.
Poruzumění mluvené řeči (Spoken language understanding, SLU) a konkrétněji sémantický parsing je zásadní komponentou každé řečové aplikace. Tento článek podává přehled současného výzkumu v oblasti SLU s důrazem na metody strojového učení používané k tomuto účelu. Sledujeme aktuální trendy v sémantickém parsingu a diskusi uzavíráme výhledem na nejslibnější směry výzkumu do budoucna.
[ "Ahmad Aghaebrahimian", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/4fbb0656d67a67686ebbad3023c71a59856b7ae7/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Discourse Relations in Czech
Diskurzní vztahy v češtině
This doctoral thesis is devoted to linguistic analysis of discourse relations as one of the aspects of discourse coherence. Discourse relations are semantic relations holding between propositions in a discourse (discourse arguments). The aim of the thesis is a complex description of discourse relations in Czech and its application in an annotation scheme in the Prague Dependency Treebank. The thesis is divided into three parts: The first one is focused on the theoretical description of discourse relations and on analysis of adequacy of various methodological concepts in corpus processing. The second part describes in detail the proposed scheme for the annotation of discourse relations and the process of the corpus build-up including the evaluation of consistency of the annotated data. Finally, in the last part of the thesis, we address some problematic issues arisen with the employment the proposed scheme and look for their possible solutions.
Doktorská práce se zabývá lingvistickou analýzou diskurzních vztahů jakožto jednoho z aspektů textové koherence. Diskurzními vztahy rozumíme významové vztahy mezi jednotlivými propozicemi v textu, tzv. diskurzními argumenty. Cílem práce je ucelený popis diskurzních vztahů v češtině a jeho vtělení do anotačního schématu Pražského závislostního korpusu. Práce je rozdělena do tří částí: První z nich je zaměřena na teoretický popis diskurzních vztahů a rozbor vhodnosti různých metodologických postupů při korpusovém zpracování. Druhá část podrobně popisuje navržené schéma pro anotaci diskurzních vztahů a proces vzniku takto značeného korpusu včetně evaluace konzistence značených dat. V poslední části práce se pak věnujeme některým problematickým okruhům při užití navrženého schématu a jejich řešení.
[ "Lucie Poláková" ]
https://www.semanticscholar.org/paper/7c9933c968c039adaf71b349ef93589a70fb66ff/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Signals of Attribution in the Prague Dependency Treebank
Příznaky atribuce v Pražském závislostním korpusu
The paper aims at mining a richly annotated treebank for features relevant in automatic annotation/detection of attribution – ascription of texts contents to agents who expressed them. Resolving attribution is an important component for representing and modeling discourse structure, but it is also an essential task in opinion mining and sentiment analysis (e.g. identification of sources).
Článek si klade za cíl nalézt v bohatě anotovaném korpusu příznaky přínosné pro automatickou anotaci/detekci atribuce – přiřazení obsahu textu zdrojům, které jej vyjadřují. Určení atribuce je významnou součástí reprezentace a modelování dirkurzních vztahů, ale je rovněž zásadní součástí např. analýzy sentimentu (identifikace zdrojů).
[ "Lucie Poláková", "Pavlína Jínová", "Jiří Mírovský" ]
https://www.semanticscholar.org/paper/c43b1515a5b57d8505edc63a86ed2afbf5d70063/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Resource-Light Acquisition of Inflectional Paradigms
Učení se paradigmat nenáročné na zdroje
This paper presents a resource-light acquisition of morphological paradigms and lexicon for fusional languages. It builds upon Paramor [10], an unsupervised system, by extending it: (1) to accept a small seed of manually provided word inflections with marked morpheme boundary; (2) to handle basic allomorphic changes acquiring the rules from the seed and/or from previously acquired paradigms. The algorithm has been tested on Czech and Slovene tagged corpora and has shown increased F-measure in comparison with the Paramor baseline.
Článek popisuje algoritmus na naučení se morfologických paradigmat a slovníku pro flektivní jazyky, který je nenáročný na zdroje.
[ "Radoslav Klíč", "Jirka Hana" ]
https://www.semanticscholar.org/paper/a72ac58d927f7d47ba56f6bec7b16946ac619dd9/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Universal Dependencies 1.1
Universal Dependencies 1.1
Universal Dependencies is a project that seeks to develop cross-linguistically consistent treebank annotation for many languages, with the goal of facilitating multilingual parser development, cross-lingual learning, and parsing research from a language typology perspective. The annotation scheme is based on (universal) Stanford dependencies (de Marneffe et al., 2006, 2008, 2014), Google universal part-of-speech tags (Petrov et al., 2012), and the Interset interlingua for morphosyntactic tagsets (Zeman, 2008). This is the second release of UD Treebanks, Version 1.1.
Universal Dependencies je projekt, který vyvíjí mezijazykově konzistentní morfologicko-syntaktickou anotaci mnoha jazyků s cílem usnadnit vývoj jazykově nezávislých parserů, mezijazykové strojové učení a výzkum parsingu z perspektivy jazykové typologie. Anotační schéma je založeno na univerzálních Stanfordských závislostech (de Marneffe et al., 2006, 2008, 2014), univerzálních značkách Google (Petrov et al., 2012) a na Intersetu, tj. interlingvě pro sady morfologických značek (Zeman, 2008). Toto je druhé vydání treebanků UD, verze 1.1.
[ "Željko Agić", "Maria Jesus Aranzabe", "Aitziber Atutxa", "Cristina Bosco", "Jinho Choi", "Marie-Catherine de Marneffe", "Timothy Dozat", "Richárd Farkas", "Jennifer Foster", "Filip Ginter", "Iakes Goenaga", "Koldo Gojenola", "Yoav Goldberg", "Jan Hajič", "Anders Johannsen", "Jenna Kanerva", "Juha Kuokkala", "Veronika Laippala", "Alessandro Lenci", "Krister Lindén", "Nikola Ljubešić", "Teresa Lynn", "Christopher Manning", "Héctor Martínez Alonso", "Ryan McDonald", "Anna Missilä", "Simonetta Montemagni", "Joakim Nivre", "Hanna Nurmi", "Petya Osenova", "Slav Petrov", "Jussi Piitulainen", "Barbara Plank", "Prokopis Prokopidis", "Sampo Pyysalo", "Wolfgang Seeker", "Mojgan Seraji", "Natalia Silveira", "Maria Simi", "Kiril Simov", "Aaron Smith", "Reut Tsarfaty", "Veronika Vincze", "Daniel Zeman" ]
https://www.semanticscholar.org/paper/1e105a47662affe083eaf110e5590ad4b9f7894d/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
A Machine Learning Approach to Hypothesis Decoding in Scene Text Recognition
Využití strojového učení při dekódování hypotéz v rozpoznávání textu z fotografií
Scene Text Recognition (STR) is a task of localizing and transcribing textual information captured in real-word images. With its increasing accuracy, it becomes a new source of textual data for standard Natural Language Processing tasks and poses new problems because of the specific nature of Scene Text. In this paper, we learn a string hypotheses decoding procedure in an STR pipeline using structured prediction methods that proved to be useful in automatic Speech Recognition and Machine Translation. The model allow to employ a wide range of typographical and language features into the decoding process. The proposed method is evaluated on a standard dataset and improves both character and word recognition performance over the baseline.
Cílem Rozpoznávání textu z fotografií (STR) je právně lokalizovat a přespat text zachycený na fotografii z reálného prostředí. Rostoucí úspěšnost rozpoznávání zároveň dělá z těchto textů zajímavý zdroj dat pro zpracování přirozeného jazyka a zároveň přináší nové problémy, které jsou specifické právě pro texty, které se na fotografiích vyskytují. V tomto článku představujeme učení dekódování textových řetězců v systému STR pomocí metod strukturní predikce, které se využívají při dekódování v rozpoznávání řeči a strojovém překladu. Model při učení využívá jazykové a typografické rysy. Navržená metoda je evaluována na standardní datové sadě a zvyšuje úspěšnost rozpoznávání znaků i rozpoznávání celých slov.
[ "Jindřich Libovický", "Lukáš Neumann", "Pavel Pecina", "Jiří Matas" ]
https://www.semanticscholar.org/paper/7e520f185014ff9a88112a52f960c452b04a08ba/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Linguistic Issues in Machine Translation between Czech and Russian
Lingvistické otázky ve strojovém překladu mezi češtinou a ruštinou
In this thesis we analyze machine translation between Czech and Russian languages from the perspective of a linguist. We explore the output of our two experimental systems and two commercial systems: PC Translator and Google Translate. We make a linguistically-motivated classification of errors for the language pair and describe each type of error in detail, analyzing whether it occurred due to some difference between Czech and Russian or is it caused by the system architecture. In particular, we focus on one specific error type - surface valency.
V této disertační práci zkoumáme strojový překlad mezi češtinou a ruštinou z hlediska lingvisty. Hlavním cílem práce je lingvistický rozbor chyb ve výstupu čtyř systémů strojového překladu, dvou experimentálních - TectoMT, Moses - a dvou komerčních - PC Translator a Google Translate. Analyzujeme každý typ chyb a řešíme, zda daná chyba souvisí s rozdílem mezi češtinou a ruštinou nebo zda je zapříčiněná architecturou jednotlivých systémů. Ve zvláštní kapitole se zaměřujeme na chyby v povrchové valenci sloves.
[ "Natalia Klyueva" ]
https://www.semanticscholar.org/paper/378121ec481c48818954f6025acd7f6bc6a26004/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Analysis of MultiWord Expression translation errors in Statistical Machine Translation
Analýza chyb VV ve statistickém strojovém překladu
In this paper, we analyse the usage of multiword expressions in Statistical Machine Translation. We exploit the Moses SMT toolkit to train models for French-English and Czech-Russian language pairs. For each language pair, two models were built: a baseline model without additional MWE data and the model enhanced with information on MWE. For the French-English pair, we tried three methods of introducing the MWE data. For Czech-Russian pair, we used just one method - adding automatically extracted data as a parallel corpus.
Zkoumáme chyby ve VV v statistickém strojovém překladu pro několik jazyků. Poživáme několik metod pro integraci VV do překladového systému.
[ "Natalia Klyueva", "Jeevanthi Liyanapathirana" ]
https://www.semanticscholar.org/paper/1c77dbdb45a1e728111f86d3b3017d4ea621e1ff/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Knowledge-based Dialog
Dialog založený na znalostech
We present a novel knowledge based dialog system that learns from interactions with users. It asks users for advice about unknown questions and generalizes them according to its knowledge base. The dialog system was evaluated through the CrowdFlower crowdsourcing platform and it showed the ability to adapt without any prior information except the knowledge base.
Představujeme nový dialogový systém učící se z interakce s uživateli. Systém žádá uživatele o radu u neznámých otázek a zobecňuje je v souladu se svou znalostní bází. Dialogový systém je vyhodnocen prostředníctvím platformy CrowdFlower.
[ "Miroslav Vodolán" ]
https://www.semanticscholar.org/paper/4f6ac483a18a9a7dcbd50fedc329005c8768e1d1/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Hybrid Dialog State Tracker
Hybridní State Tracker pro Dialog
This paper presents a hybrid dialog state tracker that combines a rule based and a machine learning based approach to belief state tracking. Therefore, we call it a hybrid tracker. The machine learning in our tracker is realized by a Long Short Term Memory (LSTM) network. To our knowledge, our hybrid tracker sets a new state-of-the-art result for the Dialog State Tracking Challenge (DSTC) 2 dataset when the system uses only live SLU as its input.
Tento článek představuje hybridní dialog state tracker, který kombinuje přístup pravidlový s přístupem strojového učení pro trackování stavu dialogu. Strojové učení je v našem trackeru realizováno s využitím Long Short Term Memory (LSTM) sítě. Podle našich současných znalostí hybridní tracker dosahuje state-of-the-art výsledků v soutěži Dialog State Tracking Challenge (DSTC) 2, v kategorii systémů využívající pouze originální SLU jako svůj vstup.
[ "Miroslav Vodolán", "Rudolf Kadlec", "Jan Kleindienst" ]
https://www.semanticscholar.org/paper/049ecf2717e8346cdc03bba55cd3e76dd09b1e18/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
CLARA: A New Generation of Researchers in Common Language Resources and Their Applications
CLARA: Výzkum nové generace v oblasti jazykových zdrojů a jejich aplikací
CLARA (Common Language Resources and Their Applications) is a Marie Curie Initial Training Network which ran from 2009 until 2014 with the aim of providing researcher training in crucial areas related to language resources and infrastructure. The scope of the project was broad and included infrastructure design, lexical semantic modeling, domain modeling, multimedia and multimodal communication, applications, and parsing technologies and grammar models. An international consortium of 9 partners and 12 associate partners employed researchers in 19 new positions and organized a training program consisting of 10 thematic courses and summer/winter schools. The project has resulted in new theoretical insights as well as new resources and tools. Most importantly, the project has trained a new generation of researchers who can perform advanced research and development in language resources and technologies.
CLARA (Common Language Resources and Their Applications) je projekt v rámci Marie Curie Initial Training Network, který probíhal v letech 2009-2014 a jehož cílem bylo poskytnout vědeckou přípravu mladým vědcům v oblasti jazykových zdrojů a zpracování jazykových dat (např. návrhy jazykových infrastruktur, lexikální sémantické modelování, modelování domény, multimediální a multimodální komunikace, technologie automatického zpracování jazyka a jejich aplikace). Projekt vyústil v nový teoretický vhled do problematiky a jeho výstupem jsou nové jazykové zdroje a nástroje, především však připravil novou generaci výzkumníků, kteří jsou schopni dále obor rozvíjet.
[ "Koenraad De Smedt", "Erhard Hinrichs", "Detmar Meurers", "Inguna Skadiņa", "Bolette Sandford Pedersen ", "Costanza Navarretta", "Núria Bel", "Krister Lindén", "Markéta Lopatková", "Jan Hajič", "Gisle Andersen", "Przemysław Lenkiewicz" ]
https://www.semanticscholar.org/paper/c955a20d79ae4c6dc91e3919ace0ec721470a077/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Building an Error-tagged Learner Corpus of Czech
Kompilace chybově anotovaného žákovského korpusu češtiny
Czesl is a corpus of texts produced by non‐native speakers of Czech. To  adequately  annotate the  deviations  in  Czech word‐order  and  a fairly complex inflectional morphology, the corpus uses two tiers of annotation,  each  tier  correcting  different  types  of  errors. Links between the tiers allow capturing errors in word order and complex discontinuous  expressions.  Errors  are  not  only  corrected,  but  also classified. We combine (1) a grammar-based taxonomy of errors in spelling,  morphology,  morphosyntax,  lexicon  and  style,  and  (2)  a formal  error  classification based on surface alternations. The annotation scheme  was  tested  on  a  data set  of  approx.  175,000 words  with  fair  inter‐annotator  agreement  results.  In  addition  to presenting  the  current state  of  the  corpus,  I  will  discuss  various decisions (design, workflow, technical) made during its creation and revisit them with the advantage of hindsight.
Přednáška prezentuje český žákovský korpus Czesl, jeho anotační schéma a způsob anotace.
[ "Jirka Hana" ]
https://www.semanticscholar.org/paper/b325505216ffa4e9c458ce7138259ee115b54997/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Sentence diagrams: their evaluation and combination
Školní větné rozbory: jejich vyhodnocení a kombinace
The purpose of our work is to explore the possibility of using sentence diagrams produced by schoolchildren as training data for automatic syntactic analysis. We have implemented a sentence diagram editor that schoolchildren can use to practice morphology and syntax. We collect their diagrams, combine them into a single diagram for each sentence and transform them into a form suitable for training a particular syntactic parser. In this study, the object language is Czech, where sentence diagrams are part of elementary school curriculum, and the target format is the annotation scheme of the Prague Dependency Treebank. We mainly focus on the evaluation of individual diagrams and on their combination into a merged better version.
Cílem naší práce je prozkoumat možnost využití školních rozborů jako trénovací data pro automatickou syntaktickou analýzu. Implementujeme editor pro procvičování tvaroslovných a větných rozborů. Následně tyto rozbory shromažďujeme a transformujeme do tvaru potřebného pro trénování vybraného parseru. Pilotním jazykem je čeština a cílovým formátem jsou anotace ve stylu Pražského závislostního korpusu. Nyní se zaměřujeme na vyhodnocení větných rozborů a jejich kombinaci s cílem získat kvalitnější rozbory.
[ "Jirka Hana", "Barbora Hladká", "Ivana Lukšová" ]
https://www.semanticscholar.org/paper/f6542bffb04ca6257b32af8556c9d366bfdea43c/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Building an Error-tagged Learner Corpus of Czech
Kompilace chybově anotovaného žákovského korpusu češtiny
Czesl is a corpus of texts produced by non‐native speakers of Czech. To adequately annotate the deviations in Czech word‐order and a fairly complex inflectional morphology, the corpus uses two tiers of annotation, each tier correcting different types of errors. Links between the tiers allow capturing errors in word order and complex discontinuous expressions. Errors are not only corrected, but also classified. We combine (1) a grammar-based taxonomy of errors in spelling, morphology, morphosyntax, lexicon and style, and (2) a formal error classification based on surface alternations. The annotation scheme was tested on a data set of approx. 175,000 words with fair inter‐annotator agreement results.
Přednáška prezentuje český žákovský korpus Czesl, jeho anotační schéma a způsob anotace.
[ "Jirka Hana", "Alexandr Rosen", "Barbora Štindlová" ]
https://www.semanticscholar.org/paper/b325505216ffa4e9c458ce7138259ee115b54997/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Building a learner corpus
Kterak vyrobit žákovský korpus
The need for data about the acquisition of Czech by non-native learners prompted the compilation of the first learner corpus of Czech. After introducing its basic design and parameters, including a multi-tier manual annotation scheme and error taxonomy, we focus on the more technical aspects: transcription of hand-written source texts, process of annotation, and options for exploiting the result, together with tools used for these tasks and decisions behind the choices. To support or even substitute manual annotation we assign some error tags automatically and use automatic annotation tools (tagger, spell checker).
Potřeba dat o akvizici češtiny cizinci vedla k vytvoření prvního žákovského korpusu češtiny. Po představení jeho základního designu a parametrů, se zaměřujeme na technické aspekty: přepis ručně psaných textů, proces anotace a možnosti využití výsledků, spolu s nástroji používanými pro tyto úkoly.
[ "Jirka Hana", "Alexandr Rosen", "Barbora Štindlová", "Jan Štěpánek" ]
https://www.semanticscholar.org/paper/b285627b937d01b74ab82f0a056c352af4db0a35/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2014
On Word Order from the Communicative Point of View
O slovosledu z komunikačního pohledu
The presented thesis is focused on the Czech word order of contextually non-bound verbal modifications. It monitors whether there is a basic order in the contextually non-bound part of the sentence (significantly predominant in frequency) in the surface word order (cf. narodit se v Brně v roce 1950 vs. narodit se v roce 1950 v Brně; literally to be born in Brno in 1950 vs. to be born in 1950 in Brno). At the same time, we try to find out the factors influencing the word order (such as the form of modifications, their lexical expression or the effect of verbal valency). Finally, we briefly compare the word order tendencies in Czech and German. For the verification of the objectives, mainly the data from the Prague Dependency Treebank are used. The work is based on the theoretical principles of Functional Generative Description. Research results demonstrate that, at least in some cases, it is possible to detect certain general tendencies to use preferably one of two possible surface word order sequences in Czech.
Práce se zabývá popisem českého slovosledu kontextově nezapojených participantů. Sleduje, zda v povrchovém slovosledu existuje jejich základní (frekvenčně výrazně převažující) pořadí (srov. narodit se v Brně v roce 1950 vs. narodit se v roce 1950 v Brně). Zároveň se věnuje sledování faktorů, které slovosled ovlivňují (např. formě participantu, způsobu jeho lexikálního vyjádření nebo vlivu slovesné valence). V závěru krátce srovnává slovosledné tendence v češtině a v němčině. Pro ověření stanovených cílů užívá zejména data Pražského závislostního korpusu. Práce teoreticky vychází ze zásad funkčního generativního popisu. Výsledky výzkumu ukazují, že v českém povrchovém slovosledu lze alespoň v některých případech vysledovat určité obecnější tendence k jednomu ze dvou možných slovosledných pořadí.
[ "Kateřina Rysová" ]
https://www.semanticscholar.org/paper/28b8b33d24181c0db6e413491ba3bf9154640741/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
On the word order of Actor and Patient in Czech
Ke slovosledu aktoru a patientu v češtině
The paper deals with word order of inner participants (Actor and Patient) in the focus-part of Czech sentences. The analysis of the sequence of Actor and Patient reveals the criteria that may influence the arrangement of sentence participants as such.
Práce se zabývá slovosledem aktantů (aktoru a patientu) v ohniskové části českých vět. Slovosledná analýza výskytů aktoru a patientu přináší popis faktorů, které mohou ovlivnit uspořádání větných participantů ve slovosledu jako takových.
[ "Kateřina Rysová" ]
https://www.semanticscholar.org/paper/098e0c42148cd1c2e5021190cdd951a269762f1c/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Sentence information structure – contextual boundness of subordinate clauses in Czech based on the corpus analysis
Aktuální členění větné – kontextová zapojenost závislých vět v češtině na základě korpusové analýzy
The paper deals with the established scale of contextual boundness of subordinate clauses in Czech and describes the reasons why some types of categories are more often presented as known then others by the Czech writers.
Poster představil tabulku kontextové zapojenosti závislých vět v češtině a popisuje důvody, proč bývají některé typy závislých vět českými pisateli častěji prezentovány jako čtenáři známé.
[ "Kateřina Rysová" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
cs
2014
Looking Back on the Jubilee 40th Year of the Olympics in the Czech Language
Ohlédnutí za jubilejním 40. ročníkem Olympiády v českém jazyce
The article describes the 40th year of the Olympics in the Czech language – it provides an commented list of tasks and evaluate the success of investigators.
Článek popisuje 40. ročník Olympiády v českém jazyce – přináší komentovaný přehled úkolů a hodnotí úspěšnost řešitelů.
[ "Kateřina Rysová", "Karel Oliva" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Valency and Word Order in Czech ― A Corpus Probe
Valence a slovosled v češtině – korpusová sonda
The paper tests a hypothesis that obligatory adverbials (in terms of the valency) follow the non-obligatory (i.e. optional) ones in the surface word order. Neither of the used methods has proved the given hypothesis but according to the results, there are several other features that influence word order of contextually non-bound free modifiers of a verb in Czech, namely position of the sentence in the text, form and length of the verb modifiers (the whole subtrees), and the semantic dependency relation (functor) of the modifiers.
V příspěvku je testována hypotéza, že valenčně obligatorní volná slovesná doplnění v povrchovém slovosledu následují za neobligatorními (fakultativními). Užité metody hypotézu nepotvrdily, ale výsledky výzkumu ukázaly, že slovosled kontextově nezapojených volných slovesných doplnění v češtině ovlivňují jiné faktory – pozice věty v textu, forma a délka slovesného doplnění a jeho syntakticko-sémantická role (funktor).
[ "Kateřina Rysová", "Jiří Mírovský" ]
https://www.semanticscholar.org/paper/22f81fe113c99aabca39a5296cebed787bc50621/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Sentence Structure and Discourse Structure (Possible parallels)
Struktura věty a diskurzu (možné souvislosti)
The present contribution is an updated and extended version of the paper from the conference DepLing 2011. It represents the first step in comparing the nature of syntactico-semantic relations present in the sentence structure to their equivalents in the discourse structure. The study is carried out on the basis of a Czech manually annotated material collected in the Prague Dependency Treebank (PDT). According to the analysis of the underlying syntactic structure of a sentence (tectogrammatics) in the PDT, we distinguish various types of relations that can be expressed both within a single sentence (i.e. in a tree) and in a larger text, beyond the sentence boundary (between trees). We suggest that, on the one hand, semantic nature of each type of these relations corresponds both within a sentence and in a larger text (i.e. a causal relation remains a causal relation) but, on the other hand, according to the semantic properties of the relations, their distribution in a sentence or between sentences is very diverse. In this study, this observation is analyzed in detail for three cases (relations of condition, specification and opposition) and further supported by similar behaviour of the English data from the Penn Discourse Treebank.
Článek je aktualizovanou a rozšířenou verzí článku z konference DepLing 2011. Reprezentuje první krok v porovnání vlastností syntakticko-sémantických vztahů přítomných ve struktuře věty a jejich ekvivalentů ve struktuře diskurzu. Studie byla provedena na ručně anotovaných datech Pražského závislostního korpusu (PDT). Podle analýzy podkladové syntaktické struktury věty (tektogramatiky) v PDT rozlišujeme řadu typů vztahů, které mohou být vyjádřeny jak v jedné větě (tj. ve stromu), tak i v delším textu, napříč hranicemi vět (mezi stromy). Tvrdíme, že tyto vztahy na jednu stranu zachovávají své sémantické vlastnosti jak uvnitř věty, tak v delším textu (např. příčinný vztah zůstává příčinným vztahem), na druhou stranu ale v souladu se sémantickými vlastnostmi vztahů je jejich distribuce uvnitř vět či mezi větami velice rozličná. V této studii toto pozorování ověřujeme na třech případech (na podmínkovém, specifikačním a opozičním vztahu) a dokládáme podobným chováním anglických dat projektu Penn Discourse Treebank.
[ "Pavlína Jínová", "Lucie Poláková", "Jiří Mírovský" ]
https://www.semanticscholar.org/paper/651af0a306f2f9d814aa62cdda904a24b06e7f66/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2014
The 25th European Summer School in Logic, Language and Information (ESSLLI 2013)
25. evropská letní škola jazyka, logiky a informace (ESSLLI 2013)
A report from the 25th European Summer School in Logic, Language and Information in Düsseldorf.
Zpráva z 25. ročníku evropské letní školy logiky, jazyka a informace v Düsseldorfu.
[ "Kateřina Veselovská" ]
https://www.semanticscholar.org/paper/391fdef53cc31b9fab1abd5f54ccd4b1fb2604c4/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2014
@#%$!!: Swear words in today´s online communication
@#%$!!: Vulgarismy v současné online komunikaci
This talk is focused on automatioc swear words detection in sentiment analysis. We outline swear words categorization and describe basic strategies when creating new vulgarisms within online communication.
Příspěvek je věnován automatické detekci vulgarismů, tj. význačných indikátorů negativního hodnocení, v rámci postojové analýzy. Zaměříme se na význam a kategorizaci vulgarismů a především popíšeme základní strategie při vytváření nových vulgarismů v internetové komunikaci, od přidávání nepísmenných znaků přes záměnu písmen po užívání speciálních zkratek a výrazů na pomezí argotu. Dále navrhneme základní řešení pro automatické zpracování těchto slov.
[ "Kateřina Veselovská" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Fear and Trembling: Annotating Emotions in Czech Holocaust Testimonies
Strach a chvění: Anotace emocí ve svědectvích o holokaustu
In this paper, we introduce the Visual History Archive of the USC Shoah Foundation as a multimodal data resource for sentiment analysis in Czech, but potentially all thirty three languages it contains. We take the opportunity of having both physical access to these unique data and the well-established research group on sentiment analysis at Charles University in Prague. Our aim is to provide methodology for sentiment annotation of these multimodal data combining subjectivity detection within a treebank with spoken term detection.
V článku představujeme Archiv vizuální historie MALACH jako multimodální zdroj dat pro postojovou analýzu v češtině, případně v dalších jazycích, které archiv obsahuje.
[ "Kateřina Veselovská" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Idiomaticity and Language Complexity: The Case of Czech and English
Idiomaticita a jazyková komplexita v porovnání češtiny a angličtiny
This survey serves as an illustration of equalization tendency in typologically different languages, using constructional analysis of structures containing adcollocational verbal idioms.
Tento výzkum rozvíjí myšlenku ekvikomplexity typologicky odlišných jazyků za pomocí konstrukční analýzy struktur obsahujících adkolokační verbální idiomy.
[ "Kateřina Veselovská" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
On Linguistic Structure of Evaluative Meaning
K lingvistické struktuře evaluativního významu
In this paper we present a preliminary research into the linguistic structure of evaluative meaning in Czech. We describe the ways the evaluative meaning is modelled in Czech using means from different layers of linguistic description, mainly morphology. Moreover, we use the construction grammar framework (see Fried and Östman 2004) to capture evaluative sentences and to depict the relationship between structure, meaning and use of evaluative expressions in language, joining the growing body of constructional research concerning the expressions of subjective judgement, broadly defined (e.g. Matsumoto 2008, Fried and Östman 2009, Terkourafi 2010).
Předběžný výzkum struktury evaluativního významu v češtině. Popisujeme způsoby, jakými je evaluativní význam modelován na různých úrovních popisu jazyka, především však na rovině morfologické. Při formalizaci využíváme konstrukčně-gramatického formalismu.
[ "Kateřina Veselovská" ]
https://www.semanticscholar.org/paper/87a0b87c3a44ad07d4392d166f6ceb340855895f/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2014
Sentiment analysis in Czech – state of the art and future work
Postojová analýza v češtině – současný stav a výhledy
In this talk we focus on sentiment analysis, i.e. linguistic analysis of emotions and opinions expressed in speech or in text.
V přednášce se zaměříme na problematiku postojové analýzy, tedy lingvistického zkoumání emocí, názorů a postojů vyjádřených v textu nebo v řeči. Shrneme dosavadní bádání v oboru a zaměříme se na jednotlivé aspekty vyjadřování emocí: na stránku lexikální (slovník hodnotících výrazů pro češtinu, evaluativní idiomy), gramatickou (význam jednotlivých slovních druhů, typické syntaktické vzorce hodnotících vět) a především sémantickou a pragmatickou. Dále představíme dostupné aplikace vycházející z postojové analýzy a nastíníme problémy a výzvy, které jsou v této disciplíně aktuální – např. automatická detekce cílů hodnocení či zapojení nadvětné syntaxe.
[ "Kateřina Veselovská" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
cs
2014
Sister's husband was a friend´s brother – On syntactic position of non-conguent attribute expressed by a noun in genitive
Sestry manžel byl kamaráda bratr – ke slovoslednému postavení neshodného atributu vyjádřeného substantivem v genitivu
The talk was dedicated to syntactic position of non-conguent attribute expressed by a noun in genitive, including categorization based on PDT data.
Příspěvek byl věnován problematice neshodného genitivního přívlastku, konkrétně možnostem jeho pronikání do antepozice. V hlavní části jsme se věnovali kategorizaci neshodných genitivních přívlastků vyskytujících se v antepozici na datech z PDT.
[ "Kateřina Veselovská" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Opinion Target Identification Using Thematic Concentration of the Text
Automatická detekce hodnocené entity za využití tematické koncentrace textu
This contribution describes improvements of opinion target identification in sentiment analysis, employing methods from quantitative linguistics, namely thematic concentration of the text.
Příspěvek popisuje možnosti vylepšení automatické detekce hodnocených entit v rámci postojové analýzy za využití metod kvantitativní lingvistiky, přesněji tematické koncentrace textu.
[ "Kateřina Veselovská", "Radek Čech" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Subjectivity Lexicon for Czech: Implementation and Improvements
Slovník českých hodnotících výrazů: implementace a vylepšování
The aim of this paper is to introduce the Czech subjectivity lexicon, a new lexical resource for sentiment analysis in Czech. We describe particular stages of the manual refinement of the lexicon and demonstrate its use in the state-of-the-art polarity classifiers, namely the Maximum Entrophy classifier. We test the success rate of the system enriched with the dictionary on different data sets, compare the results and suggest some further improvements of the lexicon-based classification system.
Cílem tohoto článku je představit Czech subjectivity lexicon, nový lexikální zdroj pro analýzu sentimentu v češtině. Popisujeme jednotlivé fáze manuálních úprav slovníku a demonstrujeme jeho využití při klasifikaci polarity, respektive pro trénování Maximum Entrophy klasifikátoru. Úspěšnost klasifikátoru obohaceného o slovník testujeme na několika datasetech. Na základě výsledků navrhujeme vylepšení stávajícího systému.
[ "Kateřina Veselovská", "Jan Hajič, jr.", "Jana Šindlerová" ]
https://www.semanticscholar.org/paper/ed57a0eed6a445280edaec85fea1b575a3b4278e/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Sentiment Detection and Annotation in a Treebank
Detekce a anotace hodnotících struktur v závislostních datech
Although sentiment analysis has been gaining a booming interest recently, most experiments use only plain text corpora. This poster shows that an existing richly annotated Prague Dependency Treebank can be exploited as a source of valuable information important for various opinion mining subtasks.
Přestože je postojová analýza široce zkoumaným odvětvím, většina výzkumu probíhá na prostém textu. Tento poster ukazuje využití závislostních dat z PDT pro nejrůznější úlohy v rámci opinion miningu.
[ "Kateřina Veselovská", "Jan Mašek", "Vladislav Kuboň" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Patterns of Evaluative Meaning: The Case of Opinion Target
Vzorce evaluativního významu: Identifikace postojového cíle
In this paper, we describe and generalize syntactic patterns of structures containing opinion targets, i.e. evaluated entities, taking into account their semantic properties.
V příspěvku popisujeme generalizované vzorce výskytu postojových cílů (opinion targets) v evaluativních konstrukcích.
[ "Kateřina Veselovská", "Jana Šindlerová" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
ÚFAL: Using Hand-crafted Rules in Aspect Based Sentiment Analysis on Parsed Data
ÚFAL: Využití ručně navržených pravidel v postojové analýze aspektů na závislostních datech
This paper describes our submission to SemEval 2014 Task 41 (aspect based sentiment analysis). The current work is based on the assumption that it could be advantageous to connect the subtasks into one workflow, not necessarily following their given order. We took part in all four subtasks (aspect term extraction, aspect term polarity, aspect category detection, aspect category polarity), using polarity items detection via various subjectivity lexicons and employing a rule-based system applied on dependency data. To determine aspect categories, we simply look up their WordNet hypernyms. For such a basic method using no machine learning techniques, we consider the results rather satisfactory.
Článek popisuje systém pro automatickou extrakci hodnocených cílů, se kterým jsme se zúčastnili soutěže SemEval 2014. Systém je pravidlový, bez zapojení strojového učení, přesto však dosahuje relativně dobrých výsledků.
[ "Kateřina Veselovská", "Aleš Tamchyna" ]
https://www.semanticscholar.org/paper/95c7ddafe6411f15658e13884ed64838723e5fa8/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Dealing with Function Words in Unsupervised Dependency Parsing
Používání funkčních slov v neřízeném závislostním parsingu
In this paper, we show some properties of function words in dependency trees. Function words are grammatical words, such as articles, prepositions, pronouns, conjunctions, or auxiliary verbs. These words are often short and very frequent in texts and therefore many of them can be easily recognized. We formulate a hypothesis that function words tend to have a fixed number of dependents and we prove this hypothesis on treebanks. Using this hypothesis, we are able to improve unsupervised dependency parsing and outperform previously published state-of-the-art results for many languages.
Používání funkčních slov v neřízeném závislostním parsingu. Používání funkčních slov v neřízeném závislostním parsingu.
[ "David Mareček", "Zdeněk Žabokrtský" ]
https://www.semanticscholar.org/paper/f2f7e3f85ea5c7261bdf0ce57d69c642dee21333/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Alex: Bootstrapping a Spoken Dialogoue System for a New Domain by Real Users
Alex: Bootstrapping hlasového dialogového systému pro novou doménu za pomoci skutečných uživatelů
When deploying a spoken dialogue system in a new domain, one faces a situation where little to no data is available to train domain-specific statistical models. We describe our experience with bootstrapping a dialogue system for public transit and weather information in real-word deployment under public use. We proceeded incrementally, starting from a minimal system put on a toll-free telephone number to collect speech data. We were able to incorporate statistical modules trained on collected data – in-domain speech recognition language models and spoken language understanding – while simultaneously extending the domain, making use of automatically generated semantic annotation. Our approach shows that a successful system can be built with minimal effort and no in-domain data at hand.
Při nasazování dialogového systému pro novou doménu je nutné se vypořádat s nedostatkem trénovacích dat pro doménově specifické statistické modely. V tomto článku popisujeme své zkušenosti s vytvářením dialogového systému pro informace o veřejné dopravě a počasí přímo za provozu s uživateli z řad veřejnosti. Postupovali jsme inkrementálně od minimálního systému, který byl nasazen na bezplatné telefonní číslo ke sběru řečových dat. Na získaných datech jsme byli schopni natrénovat statistické modely – doménové jazykové modely pro rozpoznávání řeči a model pro porozumění jazyku, který používá automaticky generovanou sémantickou anotaci. Náš postup ukazuje, že úspěšný systém lze postavit i s minimálním úsilím a bez předem dostupných trénovacích dat pro danou doménu.
[ "Ondřej Dušek", "Ondřej Plátek", "Lukáš Žilka", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/5492ce09e493de25f4db394a037859ae637b2842/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Verbal Valency Frame Detection and Selection in Czech and English
Detekce a výběr valenčního rámce v čestině a angličtině
We present a supervised learning method for verbal valency frame detection and selection, i.e., a specific kind of word sense disambiguation for verbs based on subcategorization information, which amounts to detecting mentions of events in text. We use the rich dependency annotation present in the Prague Dependency Treebanks for Czech and English, taking advantage of several analysis tools (taggers, parsers) developed on these datasets previously. The frame selection is based on manually created lexicons accompanying these treebanks, namely on PDT-Vallex for Czech and EngVallex for English. The results show that verbal predicate detection is easier for Czech, but in the subsequent frame selection task, better results have been achieved for English.
Prezentujeme metodu strojového učení („s učitelem“) pro detekci a výběr slovesných valenčních rámců, tj. specifický druh desambiguace významu sloves založený na informacích o subkategorizaci; odpovídá to detekci zmínek o událostech v textu. Používáme bohaté závislostní informace z Pražských závislostních korpusů češtiny a angličtiny a několika dříve vyvinutých nástrojů (taggery, parsery). Výběr rámce je založen na slovnících ručně sestavených pro tyto korpusy – PDT-Vallexu pro češtinu a EngVallexu pro angličtinu. Výsledky ukazují, že detekce predikátů je snažší v češtině, ale při výběr správného rámce jsme dosáhli lepších výsledků v angličtině.
[ "Ondřej Dušek", "Jan Hajič", "Zdeňka Urešová" ]
https://www.semanticscholar.org/paper/f5aaf4f58e62024a0544bd28163eb29bdc85abb0/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Khresmoi Summary Translation Test Data 1.1
Testovací data projektu Khresmoi pro překlad souhrnů 1.1
This package contains data sets for development and testing of machine translation of sentences from summaries of medical articles between Czech, English, French, and German.
Tento balíček obsahuje sady paralelních vět pro vývoj a testování strojového překladu souhrnů vědeckých článků z oboru medicíny mezi češtinou, angličtinou, francouzštinou a němčinou.
[ "Ondřej Dušek", "Jan Hajič", "Jaroslava Hlaváčová", "Pavel Pecina", "Aleš Tamchyna", "Zdeňka Urešová" ]
https://www.semanticscholar.org/paper/e9f2ebb12dc230f165e58047c05bbc1c2b4d9b85/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Machine Translation of Medical Texts in the Khresmoi Project
Strojový překlad lékařských textů v projektu Khresmoi
The WMT 2014 Medical Translation Task poses an interesting challenge for Machine Translation (MT). In the standard translation task, the end application is the translation itself. In this task, the MT system is considered a part of a larger system for cross-lingual information retrieval (IR).
Lékařská překladová úloha na WMT 2014 představuje zajímavou výzvu pro strojový překlad (machine translation, MT). Ve standardní překladové úloze je koncovou aplikací překlad sám o sobě. Naproti tomu v této úloze je systém strojového překladu součástí většího systému pro mezijazykové vyhledávání informací (information retrieval, IR).
[ "Ondřej Dušek", "Jan Hajič", "Jaroslava Hlaváčová", "Michal Novák", "Pavel Pecina", "Rudolf Rosa", "Aleš Tamchyna", "Zdeňka Urešová", "Daniel Zeman" ]
https://www.semanticscholar.org/paper/3e546c196864ec3f28e15b241c6acd3b44090250/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Free English and Czech telephone speech corpus shared under the CC-BY-SA 3.0 license
Anglický a český korpus telefonních nahrávek volně dostupný pod licencí CC-BY-SA 3.0
We present a dataset of telephone conversations in English and Czech, developed to train acoustic models for automatic speech recognition (ASR) in spoken dialogue systems (SDSs). The data comprise 45 hours of speech in English and over 18 hours in Czech. All audio data and a large part of transcriptions was collected using crowdsourcing; the rest was transcribed by hired transcribers. We release the data together with scripts for data re-processing and building acoustic models using the HTK and Kaldi ASR toolkits. We publish the trained models described in this paper as well. The data are released under the CC-BY-SA 3.0 license, the scripts are licensed under Apache 2.0. In the paper, we report on the methodology of collecting the data, on the size and properties of the data, and on the scripts and their use. We verify the usability of the datasets by training and valuating acoustic models using the presented data and scripts.
Prezentujeme sadu nahrávek telefonních hovorů v angličtině a češtině, vytvořenou pro trénování akustických modelů pro automatické rozpoznávání řeči v hlasových dialogových systémech. Data sestávají ze 45 hodin nahrávek v angličtině a více než 18 hodin v češtině. Všechna data a část transkripcí byla získána pomocí crowdsourcingu, zbytek byl přepsán profesionálně. Data zveřejňujeme společně se skripty pro preprocessing a sestavení akustických modelů v nástrojích HTK a Kaldi, včetně modelů natrénovaných na našich datech. Data jsou licencována pod CC-BY-SA 3.0, skripty pod licencí Apache 2.0. V článku popisujeme metodiku sběru dat, jejich velikost a vlastnosti, a trénovací skripty a jejich použití. Použitelnost dat a skriptů demostrujeme natrénováním a validací akustických modelů.
[ "Matěj Korvas", "Ondřej Plátek", "Ondřej Dušek", "Lukáš Žilka", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/d86a73c8d48dfbf28f82089a46566978eb175300/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Vystadial 2013 – ASR training scripts
Vystadial 2013 – ASR trénovací scrikripty
Vystadial 2013 ASR training scripts provides ASR training scripts for HTK or KALDI developed for training acoustic models for automatic speech recognition in spoken dialogue systems.
Vystadial 2013 ASR trénovací skripty obsahují skripty pro HTK a KALDI vyvinuté pro trénování akustických modelů pro automatické rozpoznávání řeči v dialogových systémech.
[ "Matěj Korvas", "Ondřej Plátek", "Ondřej Dušek", "Lukáš Žilka", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/be2512c4b47081baf1eaa23f7bd64ff8016c0cbc/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2014
Vystadial 2013 – Czech data
Vystadial 2013 – Česká data
Vystadial 2013 is a dataset of telephone conversations in Czech, developed for training acoustic models for automatic speech recognition in spoken dialogue systems. The data comprise over over 15 hours in Czech, plus orthographic transcriptions.
Vystadial 2013 je databáze telefonních hovorů v češtině, vyvinuté pro trénování akustických modelů pro automatické rozpoznávání řeči v dialogových systémech. Data obsahují více než 15 hodin v českém jazyce.
[ "Matěj Korvas", "Ondřej Plátek", "Ondřej Dušek", "Lukáš Žilka", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/21f368ae92dfebbe88cbd64b1ae66f87661a3338/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Vystadial 2013 – English data
Vystadial 2013 – Anglická data
Vystadial 2013 is a dataset of telephone conversations in English, developed for training acoustic models for automatic speech recognition in spoken dialogue systems. The data comprise over over 41 hours in English, plus orthographic transcriptions.
Vystadial 2013 je databáze telefonních hovorů v anglickém jazyce, vyvinuté pro trénování akustických modelů pro automatické rozpoznávání řeči v dialogových systémech. Data obsahují více než 41 hodin v anglickém jazyce.
[ "Matěj Korvas", "Ondřej Plátek", "Ondřej Dušek", "Lukáš Žilka", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/39259de0b7aee6ab54b0b373bbeecb01e7724e95/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Czech Light Verbs from a Lexicographic Point of View
Česká "lehká" slovesa z lexikografického hlediska
Light verbs – representing a type of complex predicates – pose a serious challenge for both theoretical and applied linguistics as their syntactic structures are determined not solely by verbs alone but also by predicative elements with which light verbs combine. In this contribution, syntactic and semantic properties of Czech light verbs combined with predicative nouns are discussed and an adequate and economical lexicographic representation of these phenomena are proposed. Moreover, two tasks are addressed: compiling an inventory of Czech light verbs entering into combination with predicative nouns and adopting criteria for distinguishing light verbs from main verbs.
"Lehká" slovesa představující určitý typ komplexních predikátů představují značný problém jak pro teoretickou, tak pro komputační lingvistiku. Syntaktické struktury, které daná slovesa zakládají, jsou dány nejen slovesem, ale i predikativním elementem, se kterým se daná slovesa kombinují. V uvedeném příspěvku jsou analyzovány ze syntaktického i sémantického hlediska česká "lehká" slovesa kombinovaná s predikativními jmény. Je navržena jejich teoreticky adekvátní a zároveň úsporná lexikografická reprezentace. Dále jsou probírány zejména dva aspekty: možnost sestavení inventáře českých "lehkých" sloves a kritéria pro jejich rozpoznání.
[ "Václava Kettnerová" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
cs
2014
Lexical semantic conversions in a valency lexicon
Lexikálně-sémantické konverze ve valenčním slovníku
In this work, a formal model of a lexicographic description of selected changes in valency structure of Czech verbs, alternations, is proposed. This model is based on a thorough analysis of two types of Czech alternations. The proposed lexicographic representation supposes separate lexical units of verbs linked by a special attribute in the data component of the lexicon and general rules determining the mapping between participants and valency complementations of verbs.
Práce předkládá návrh formalizovaného lexikografického modelu alternací,tedy změn v syntaktické struktuře sloves, a jejich lingvisticky adekvátního, přitom ekonomického popisu ve slovníku. Jádrem práce je podrobný rozbor dvou vybraných typů lexikalizovaných alternací. Na základě této analýzy je navržen systém popisu alternací pro češtinu sestávající z lexikálních jednotek propojených atributy určujícími typy možných alternací (datová komponenta slovníku) a z obecných pravidel popisujících korespondenci situačních participantů a valenčních doplnění (pravidlová komponenta slovníku).
[ "Václava Kettnerová" ]
https://www.semanticscholar.org/paper/52fe17da350999456e9bb9f381d3f6d23bf2f98e/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Reflexive Verbs in a Valency Lexicon: The Case of Czech Reflexive Morphemes
Reflexivní slovesa ve valenčním slovníku: Příklad českých reflexivních morfémů
Thi paper deals with Czech reflexive verbs from the lexicographic point of view. We show that the Czech reflexive morphemes se/si constitute different linguistic meanings: either they are formal means of the word formation process of the so called reflexivization, or they are associated with the syntactic phenomena of reflexivity, reciprocity, and diatheses. All of these processes are associated with changes in the valency structure of verbs. We formulate a proposal for their lexicographic representation for the valency lexicon of Czech verbs, VALLEX.
Článek pojednává o českých reflexivních slovesech z lexikografického hlediska. Ukazujeme, že reflexivní morfémy se/si plní v češtině různé funkce: (i) jsou to buď prostředky slovotvorného procesu reflexivizace, nebo (ii) prostředky vlastní (syntaktické) reflexivity, reciprocity, příp. diatezí. Všechny uvedené jazykové procesy jsou spoojeny se změnami ve valenční struktuře sloves. V článku podáváme návrh na jejich zachycení ve valenčním slovníku VALLEX.
[ "Václava Kettnerová", "Markéta Lopatková" ]
https://www.semanticscholar.org/paper/0453e322aa36f60c40e0a1f3035c94ca0edc8fde/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
An Interplay between Valency Information and Reflexivity
Valenční informace a reflexivita
Roland Wagner’s contribution published in the last volume of the PBML journal – focusing (among other ideas) on the role of Czech reflexives – presents several critical remarks concerning the Functional Generative Description. These remarks represent a good challenge for the authors developing this model to fill empirical gaps and to make clear some theoretical presuppositions concerning valency frames of verbs and their respective reflexive counterparts that are primarily addressed by RW’s critical survey.
Roland Wagner v článku zaměřeném na česká reflexivní slovesa publikovaném v PBML prezentuje kritické výhrady k Funkčnímu generativnímu popisu (FGP). Pro autory FGP představuje jeho článek výzvu k doplnění mezer v analýze daného jevu v rámci FGP a k vyjasnění některých teoretických předpokladů týkajících se valence sloves a jejich reflexivních protějšků.
[ "Václava Kettnerová", "Markéta Lopatková", "Jarmila Panevová" ]
https://www.semanticscholar.org/paper/fd12b11358e9ed5eac7a9042f0e66f51fc49978d/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
SemEval 2014 Task 8: Broad-Coverage Semantic Dependency Parsing
Úloha 8 na SemEval 2014: Závislostní sémantická analýza se širokým pokrytím
We define broad-coverage semantic dependency parsing as the task of recovering sentence-internal predicate-argument relationships for all content words i.e. a semantic dependency structure that constitutes the core structure of sentence meaning.
Závislostní sémantickou analýzu se širokým pokrytím definujeme jako úlohu nalézt skladební dvojice mezi všemi autosémantickými slovy ve větě, tj. strukturu sémantických závislostí, která reprezentuje jádro významu věty.
[ "Stephan Oepen", "Marco Kuhlmann", "Daniel Zeman", "Yusuke Miyao", "Dan Flickinger", "Jan Hajič", "Angelina Ivanova", "Yi Zhang" ]
https://www.semanticscholar.org/paper/d4021f9e979227da726c5b1d2620002446300dc1/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Open-Source Tools for Morphology, Lemmatization, POS Tagging and Named Entity Recognition
Open-source nástroje pro morfologii, lematizaci, rozpoznávání slovních druhů a pojmenovaných entit
We present two recently released open-source taggers: NameTag is a free software for named entity recognition (NER) which achieves state-of-the-art performance on Czech; MorphoDiTa (Morphological Dictionary and Tagger) performs morphological analysis (with lemmatization), morphological generation, tagging and tokenization with state-of-the-art results for Czech and a throughput around 10-200K words per second. The taggers can be trained for any language for which annotated data exist, but they are specifically designed to be efficient for inflective languages. Both tools are free software under LGPL license and are distributed along with trained linguistic models which are free for non-commercial use under the CC BY-NC-SA license. The releases include standalone tools, C++ libraries with Java, Python and Perl bindings and web services.
V této práci představujeme dva nedávno vydané open-source nástroje: NameTag je volně šiřitelný software pro rozpoznávání pojmenovaných entit, který dosahuje nejlepších známých výsledků na češtině; MorphoDiTa provádí morfologickou analýzu (s lematizací), morfologické generování, značkování a tokenizaci s nejlepšími známými výsledky pro češtinu a rychlostí zpracování kolem 10-200 tisíc slov za sekundu. Nástroje mohou být natrénovány pro libovolný jazyk, pro který jsou k dispozici anotovaná data, jsou však zvlášť navrženy tak, aby byly efektivní pro flexivní jazyky. Oba nástroje jsou volně šiřitelné pod licencí LGPL a jsou distribuovány spolu z předtrénovanými lingvistickými modely, které jsou zdarma pro nekomerční využití podle licence CC BY-NC-SA. Vydání zahrnují samostatné nástroje, knihovny v C++ s vazbami pro Javu, Python a Perl, a konečně webové služby.
[ "Jana Straková", "Milan Straka", "Jan Hajič" ]
https://www.semanticscholar.org/paper/75a79dc572d182937688b3d064160d78437d6792/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Votter Corpus: A Corpus of Social Polling Language
Votter Corpus: Corpus sociálního dotazovacího jazyka
The Votter Corpus is a new annotated corpus of social polling questions and answers. The Votter Corpus is novel in its use of the mobile application format and novel in its coverage of specific demographics. With over 26,000 polls and close to 1 millions votes, the Votter Corpus covers everyday question and answer language, primarily for users who are female and between the ages of 13-24. The corpus is annotated by topic and by popularity of particular answers. The corpus contains many unique characteristics such as emoticons, common mobile misspellings, and images associated with many of the questions. The corpus is a collection of questions and answers from The Votter App on the Android operating system. Data is created solely on this mobile platform which differs from most social media corpora. The Votter Corpus is being made available online in XML format for research and non-commercial use. The Votter android app can be downloaded for free in most android app stores.
Votter Corpus je nový anotovaný soubor sociálních dotazů a odpovědí. Votter Corpus je nový ve svém použití formátu mobilních aplikací a románu v jeho pokrytí specifických demografických. S více než 26 000 hlasy a téměř 1 miliony hlasů pokrývá Votter Corpus každodenní otázky a odpovědi, a to především pro uživatele, kteří jsou ve věku od 13 do 24 let.
[ "Nathan David Green", "Septina Dian Larasati" ]
https://www.semanticscholar.org/paper/966ff16bdcaa1909ba171a5d33808b6a296cec97/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Not an Interlingua, But Close: Comparison of English AMRs to Chinese and Czech
Ještě ne interlingva, ale už dost blízko: Srovnání anglických AMR s čínskými a českými
Abstract Meaning Representations (AMRs) are rooted, directional and labeled graphs that abstract away from morpho-syntactic idiosyncrasies such as word category (verbs and nouns), word order, and function words (determiners, some prepositions). Because these syntactic idiosyncrasies account for many of the cross-lingual differences, it would be interesting to see if this representation can serve, e.g., as a useful, minimally divergent transfer layer in machine translation. To answer this question, we have translated 100 English sentences that have existing AMRs into Chinese and Czech to create AMRs for them. A cross-linguistic comparison of English to Chinese and Czech AMRs reveals both cases where the AMRs for the language pairs align well structurally and cases of linguistic divergence. We found that the level of compatibility of AMR between English and Chinese is higher than between English and Czech. We believe this kind of comparison is beneficial to further refining the annotation standards for each of the three languages and will lead to more compatible annotation guidelines between the languages.
Reprezentace AMR abstrahují od morfologicko- syntaktických jevů, protože mnohé z nich mohou být důvodem mezijazykových rozdílů. V článku se snažíme zjistit, zda by AMR mohly napomoci počítačovému překladu. Využíváme k tomu porovnání překladu a anotace anglických vět do češtiny a čínštiny.
[ "Nianwen Xue", "Ondřej Bojar", "Jan Hajič", "Martha Palmer", "Zdeňka Urešová", "Xiuhong Zhang" ]
https://www.semanticscholar.org/paper/7099fd783370d14cdd6b5bc399feca970b658563/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Universal Dependencies
Univerzální závislosti
Universal Dependencies is a project that seeks to develop cross-linguistically consistent treebank annotation for many languages, with the goal of facilitating multilingual parser development, cross-lingual learning, and parsing research from a language typology perspective. The annotation scheme is based on (universal) Stanford dependencies (de Marneffe et al., 2006, 2008, 2014), Google universal part-of-speech tags (Petrov et al., 2012), and the Interset interlingua for morphosyntactic tagsets (Zeman, 2008). The general philosophy is to provide a universal inventory of categories and guidelines to facilitate consistent annotation of similar constructions across languages, while allowing language-specific extensions when necessary.
Univerzální závislosti (Universal Dependencies) je projekt, jehož cílem je vytvořit mezijazykově konzistentní závislostní anotaci pro mnoho jazyků, s cílem zjednodušit vývoj jazykově nezávislých parserů, mezijazykové učení a výzkum parsingu z pohledu jazykové typologie. Anotační schéma je založeno na (univerzálních) Stanfordských závislostech (de Marneffe et al., 2006, 2008, 2014), univerzálních značkách Google (Petrov et al., 2012) a Intersetu, interlinguy pro sady morfologických značek (Zeman, 2008).
[ "Jinho Choi", "Marie-Catherine de Marneffe", "Timothy Dozat", "Filip Ginter", "Yoav Goldberg", "Jan Hajič", "Christopher Manning", "Ryan McDonald", "Joakim Nivre", "Slav Petrov", "Sampo Pyysalo", "Natalia Silveira", "Reut Tsarfaty", "Daniel Zeman" ]
https://www.semanticscholar.org/paper/b805693c17961af2cc7f859c1a54320b26036f46/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
CUNI at MediaEval 2014 Search and Hyperlinking Task: Search Task Experiments
CUNI na MediaEval 2014, Search and Hyperlinking Task: Experimenty v úlohe Search
In this paper, we describe our participation in the Search part of the Search and Hyperlinking Task in MediaEval Benchmark 2014. In our experiments, we compare two types of segmentation: fixed-length segmentation and segmentation employing Decision Trees on various features. We also show usefulness of exploiting metadata and explore removal of overlapping retrieved segments.
V článku popisujeme našu účasť v úlohe Search v Search and Hyperlinking Task vrácmi Benchmarku MediaEval 2014. Na základe našich experimentov porovnávame dva typy segmentácie: segmentácia na úseky rovnakej dĺžky a segmentácia, ktorá využíva rozhodovacie stromy. Taktiež ukážeme užitočnosť využitia metadát a preskúmame odstraňovanie prekrývajucich sa relevantných segmentov.
[ "Petra Galuščáková", "Pavel Pecina" ]
https://www.semanticscholar.org/paper/368c05ba0af8e94d12a1bdd749b35102885f8ae3/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Experiments with Segmentation Strategies for Passage Retrieval in Audio-Visual Documents
Experimenty so segmentáciou pri vyhľadávaní segmentov v audio-vizuálnych dokumentoch
This paper deals with Information Retrieval from audio-visual recordings. Such recordings are often quite long and users may want to find the exact starting points of relevant passages they search for. In Passage Retrieval, the recordings are automatically segmented into smaller parts, on which the standard retrieval techniques are applied. In this paper, we discuss various techniques for segmentation of audio-visual recordings and focus on machine learning approaches which decide on segment boundaries based on various features combined in a decision-tree model. Our experiments are carried out on the data used for the Search and Hyperlinking Task and Similar Segments in Social Speech Task of the MediaEval Benchmark 2013.
Článok sa zaoberá vyhľadávaním informácií v audio-vizuálnych nahrávkach. Takéto nahrávky sú často dlhé a pre užívateľa môže byť preto dôležité vyhľadanie presného relevantného úseku. Pri vyhľadávaní relevantných segmentov sú nahrávky najprv rozdelené na menšie časti, na ktoré sú aplikované štandardné metódy vyhľadávania informácií. V článku porovnávame niekoľko metód, ktoré slúžia na segmentáciu nahrávok, zvlášť sa zameriavame na prístupy založené na strojovom učení.
[ "Petra Galuščáková", "Pavel Pecina" ]
https://www.semanticscholar.org/paper/f6f3780ab15168235ecbbb47dae7b3899cd8182e/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
CUNI at MediaEval 2014 Search and Hyperlinking Task: Visual and Prosodic Features in Hyperlinking
CUNI na MediaEval 2014, Search and Hyperlinking Task: Vizuálne a prozodické príznaky v úlohe Hyperlinking
In this report, we present our experiments performed for the Hyperlinking part of the Search and Hyperlinking Task in MediaEval Benchmark 2014. Our system successfully combines features from multiple modalities (textual, visual, and prosodic) and confirms the positive effect of our former method for segmentation based on Decision Trees.
V článku prezentujeme naše experimenty vrámci úlohy Hyperlinking v Search and Hyperlinking Task na Benchmarku MediaEval 2014. Náš systém úspešne kombinuje príznaky rôznych modalít (textové, vizuálne, prozodické). V článku tiež ukážeme vhodnosť použitia nami navrhnutých segmentačných metód, ktoré využívajú rozhodovacie stromy.
[ "Petra Galuščáková", "Martin Kruliš", "Jakub Lokoč", "Pavel Pecina" ]
https://www.semanticscholar.org/paper/1a0d51b07ac326ee94fc6e3a8ceadd4335076cba/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
In-House: An Ensemble of Pre-Existing Off-the-Shelf Parsers
In-House: Soubor dříve existujících parserů z dílny poskytovatelů dat
As a submission of its own to the open track of Task 8 at SemEval 2014, a subset of the organizers sought to connect the task to pre-existing, ‘in-house’ parsing systems that can output the same types of semantic dependency graphs as used in the task. For each of the three target formats, there is an existing parsing system, which typically was developed in parallel to the creation of the target dependency graphs, viz. (a) for the DM format, the parser using the hand-engineered English Resource Grammar; (b) for the PAS format, the Enju parsing system, with its probabilistic HPSG acquired through linguistic projection of the PTB; and (c) for the PCEDT format, the scenario for English tectogrammatical analysis within the Treex framework, comprising a pipeline of a syntactical data-driven dependency parser and a number of hand-engineered processing blocks that convert trees from the analytical to the tectogrammatical layer.
Podmnožina organizátorů Úlohy 8 na SemEvalu 2014 se pokusila použít k řešení úlohy parsery, které byly již dříve vyvinuty pro jednotlivé datové formáty využité v této úloze. Kombinace výstupů těchto parserů byla zařazena jako samostatné řešení otevřené části úlohy (open track). Použité systémy byly typicky vyvíjeny souběžně s anotací dat, konkrétně (a) pro formát DM jde o parser nad ručně sestavenou English Resource Grammar; (b) pro formát PAS jde o systém Enju s pravděpodobnostní HPSG, získanou lingvistickou projekcí PTB; a (c) pro formát PCEDT jde o scénář anglické tektogramatické analýzy v prostředí Treex, zahrnující statistický závislostní analyzátor a řadu cílených zpracovacích bloků, které převádějí stromy z analytické na tektogramatickou rovinu.
[ "Yusuke Miyao", "Stephan Oepen", "Daniel Zeman" ]
https://www.semanticscholar.org/paper/e885fa88e3bd7d25f7420104aaf8ae5d3c8915c9/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
A Tagged Corpus and a Tagger for Urdu
Otagovaný korpus a tagger pro urdštinu
In this paper, we describe a release of a sizeable monolingual Urdu corpus automatically tagged with part-of-speech tags. We extend the work of Jawaid and Bojar (2012) who use three different taggers and then apply a voting scheme to disambiguate among the different choices suggested by each tagger. We run this complex ensemble on a large monolingual corpus and release the tagged corpus. Additionally, we use this data to train a single standalone tagger which will hopefully significantly simplify Urdu processing. The standalone tagger obtains the accuracy of 88.74% on test data.
V článku popisujeme vydání rozsáhlého jednojazyčného korpusu urdštiny s automatickým značkováním slovních druhů. Navazujeme na práci Jawaid a Bojar (2012), kde byly pro značkování použity tři taggery a finální výsledek určilo jejich hlasování. Používáme stejnou komplexní sestavu na velký jednojazyčný korpus a výsledek zpřístupňujeme veřejnosti. Kromě toho na tomto velkém korpusu trénujeme jeden samostatný tagger, což, doufáme, podstatě zjednoduší zpracování urdštiny. Tento samostatný tagger na nezávislých testovacích datech dosahuje přenosti 88,74 %.
[ "Bushra Jawaid", "Amir Kamran", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/af3c61e40f1f5871e76c3a344dff9382b3b2d4b4/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
English to Urdu Statistical Machine Translation: Establishing a Baseline
Angličtina na Urdu statistického strojového překladu:
The aim of this paper is to categorize and present the existence of resources for English- to-Urdu machine translation (MT) and to establish an empirical baseline for this task. By doing so, we hope to set up a common ground for MT research with Urdu to allow for a congruent progress in this field. We build baseline phrase-based MT (PBMT) and hierarchical MT systems and report the results on 3 official independent test sets. On all test sets, hierarchial MT significantly outperformed PBMT. The highest single-reference BLEU score is achieved by the hierarchical system and reaches 21.58% but this figure depends on the randomly selected test set. Our manual evaluation of 175 sentences suggests that in 45% of sentences, the hierarchical MT is ranked better than the PBMT output compared to 21% of sentences where PBMT wins, the rest being equal.
Článek představuje shrnutí existujících jazykových zdrojů pro překlad z angličtiny do urdštiny a demonstruje výsledky dosažené frázovým a hierarchickým statistickým překladem na třech oficiálních testovacích sadách vět. Ve všech případech hierarchický překlad dopadá lépe.
[ "Bushra Jawaid", "Amir Kamran", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/b9a6e8587a4012a0e945ac5fe84408554d96c9eb/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Two-Step Machine Translation with Lattices
Dvoukrokový strojový překlad se svazy
The idea of two-step machine translation was introduced to divide the complexity of the search space into two independent steps: (1) lexical translation and reordering, and (2) conjugation and declination in the target language. In this paper, we extend the two-step machine translation structure by replacing state-of-the-art phrase-based machine translation with the hierarchical machine translation in the 1st step. We further extend the fixed string-based input format of the 2nd step with word lattices (Dyer et al., 2008); this provides the 2nd step with the opportunity to choose among a sample of possible reorderings instead of relying on the single best one as produced by the 1st step.
Článek navazuje na starší práce o dvoukrokovém překladu a rozšiřuje je ve dvou směrech. Jednak v prvním kroku nahrazujeme frázový překlad hierarchickým, a jednak na rozhraní mezi prvním a druhým krokem používáme svaz slov jako záměrně víceznačnou reprezetanci.
[ "Bushra Jawaid", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/ec360a945875f1c2303472d6f7ef1929f3db5989/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Semi-automatic Detection of Multiword Expressions in the Slovak Dependency Treebank
Poloautomatická detekce víceslovných výrazů ve Slovenském závislostním korpusu
We describe a method for semi-automatic extraction of Slovak multiword expressions (MWEs) from a dependency treebank. The process uses an automatic conversion from dependency syntactic trees to deep syntax and automatic tagging of verbal argument nodes based on a valency dictionary. Both the valency dictionary and the treebank conversion were adapted from the corresponding Czech versions; the automatically translated valency dictionary has been manually proofread and corrected. There are two main achievements – a valency dictionary of Slovak MWEs with direct links to corresponding expressions in the Czech dictionary, PDT-Vallex, and a method of extraction of MWEs from the Slovak Dependency Treebank. The extraction reached very high precision but lower recall in a manual evaluation. This is a work in progress, the overall goal of which is twofold: to create a Slovak language valency dictionary paralleling the Czech one, with bilingual links; and to use the extracted verbal frames in a collocation dictionary of Slovak verbs.
Popisujeme metodu poloautomatické extrakce slovenských víceslovných výrazů ze závislostního korpusu. Proces používá automatickou konverzi ze závislostních syntaktických stromů do hloubkové syntaxe a automatické značkování slovesných doplnění na základě valenčního slovníku. Jak valenční slovník tak konverze syntaktického korpusu vznikla úpravou podobného nástroje pro češtinu; autmomaticky přeložený valenční slovník byl ručně zkontrolován a opraven. Přínos této práce je dvojí – valenční slovník slovenských víceslovných výrazů s přímými odkazy na odpovídající výrazy v českém slovníku PDT-Vallex a metoda pro extrakci víceslovných výrazů ze Slovenského závislostního korpusu. Práce na projektu stále probíhá, cílem je 1) vytvořit slovenský valenční slovník paralelní k českému a 2) použít extrahované slovesné rámce v kolokačním slovníku slovenských sloves.
[ "Daniela Majchráková", "Ondřej Dušek", "Jan Hajič", "Agáta Karčová", "Radovan Garabík" ]
https://www.semanticscholar.org/paper/2edf4ed8b81ef87ba1e713da36c121619e706d25/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Centrum vizuální historie Malach
Malach Center for Visual History
Poster summarizing the basic information on Malach Center for Visual History, its activities and available archival resources.
Poster představující souhrnné informace o Centru vizuální historie Malach, jeho aktivitách a dostupných archivních zdrojích.
[ "Jakub Mlynář" ]
https://www.semanticscholar.org/paper/6b335c2e2edfcf5ff3b306e966e01d0cf1692731/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Collective Memory and Language: Insights in Social Theory
Kolektivní paměť a jazyk z perspektivy sociální teorie
Various attempts to conceptualize the often vaguely used term collective memory come to the conclusion that collective memory is deeply related to linguistic and narrative phenomena. In the present paper, I aim to provide an overview and discussion of the link between language and collective memory in the context of social theory. In the case of the founding theoretical figures, M. Halbwachs and J. Assmann, the importance of language in relation to the issues of collective memory is profound. In the past two decades, the specific role of narrative and conversation had become an important subject in researching collective memory. In empirical research, on the other hand, the relationship of language and collective memory seems to be rather underrepresented. Various fields and disciplines deal with similar topics quite differently, and they also differ in the degree of explicit scrutiny of the collective memory phenomena.
Různé pokusy o konceptualizaci často vágně užívaného pojmu kolektivní paměť dospívají k závěru, že kolektivní paměť je hluboce provázaná s jazykovými a narativními fenomény. V tomto příspěvku je mým cílem předložit přehled a diskusi souvislosti mezi jazykem a kolektivní pamětí v kontextu sociální teorie. V díle zakládajících teoretických postav M. Halbwachse a J. Assmanna je význam jazyka ve vztahu k otázkám kolektivní paměti chápán jako ústřední. Stejně tak v posledních dvou dekádách se ze specifické role narativu a konverzace stává významný předmět teoretického výzkumu kolektivní paměti. Na druhou stranu, v empirických šetřeních jako by byl vztah jazyka a kolektivní paměti spíše podceňován. Ukazuje se, že různé společenskovědní a humanitní disciplíny se s podobnými tématy vyrovnávají značně odlišně a rozdíly lze nalézt také v míře explicitní pozornosti věnované tématu kolektivní paměti.
[ "Jakub Mlynář" ]
https://www.semanticscholar.org/paper/25d0e088bfce994100eb99f96d83d4d3a46517ba/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Identity as Memory and Narrative: The narrative identifications of Czechoslovak Jewish Holocaust survivors
Identita jako paměť a narace: Narativní identifikace československých židů přeživších holocaust
The topics of memory, identity and narrative are deeply intertwined. Since John Locke (1632–1704), it is quite widely acknowledged that the very sense of one’s personal identity is based on (autobiographical) memory. During the 20th century, study of collective identity and collective memory emerged in the social sciences, along with questions about possible analogy between the mechanisms of individual identity/memory and collective identity/memory. The dialectics of memories and identities have also been connected by the notion of narrative. Therefore the identity is articulated (or even constructed) by linguistic means as a meaningful life story on the basis of past experiences and biographic work. In case of oral historical interviews, the triad of identity—memory—narrative takes on specific features. I will focus on the USC Shoah Foundation’s Visual History Archive as a data resource and try to provide answers to several questions, including: How are different social identities manifested in the audiovisual oral histories? Is there a narrative expression of plurality of identities, conflicts and tensions? Are these topics somehow present in the testimonies of the Holocaust survivors? How is the sociological point of view contributing to this kind of research? As we will see, oral history is indeed a rich data resource for researching the social aspects of collective and personal identities. However, because language is privileged in oral history, we might be missing another important means of identity expression (in non-verbal ways of individual or collective action). To a certain extent, this may be overcome due to the increasing number of videotaped oral history interviews.
Dialektika paměti a identity je ve společenských vědách často spojována s tématem narativity. Identita je artikulována (či dokonce utvářena) jazykovými prostředky jako smysluplný životní příběh na základě zapamatovaných prožitých zkušeností a biografické práce. Orální historie (konkrétně Archiv vizuální historie USC Shoah Foundation) poskytuje bohatý zdroj materiálu pro zkoumání osobní, sociální i kolektivní identifikace narátorů, ale zároveň klade výzkum identit před určitá úskalí.
[ "Jakub Mlynář" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
cs
2014
Malach as a Gateway to History through Survivors' Narratives
Malach jako brána do minulosti skrze vyprávění přeživších
The article provides readers with basic information on the Malach Center for Visual History and available archival resources in the context of oral history as a specific method of capturing the past.
Článek poskytuje základní informace o Centru vizuální historie Malach a dostupných archivních zdrojích v kontextu orální historie jako specifické metody zachycení minulosti.
[ "Jakub Mlynář" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Sedimenting Past: Towards the Sociology of Oral History
Sedimentace minulosti: K sociologii orální historie
Oral history - whether as a qualitative method or discipline *sui generis* - is indeed receiving growing attention in the past decades, from scholars as well as NGO's and also the wide public. Many different oral historical projects are being conducted, in varying contexts and regions, focusing on a variety of historical events, issues and research questions, with different budgets and technological backgrounds. From a sociological point of view, it is possible to interpret the universal practice of oral history as a social phenomenon of its own kind. Sociology can approach oral history from two different points of view, which are also shaping the basic structure of my talk: oral history (1) *as a subject itself* - the sociological interpretation of oral history as a social practice; (2) *as a secondary data source* - the sociological interpretation of existing oral historical data. First, I will present some approaches to oral history as a subject of sociological research, consisting of two distinctive but complementary dimensions: (1.1) *micro-social*, and (1.2) *macro-social*. The traditional (and tricky) antinomy of individual and society is also manifesting itself in oral history, but certainly with some specific features. I will elaborate on the possible facilitation of oral history in sociological research, and also on the limitations of it. The central argument is that sociological insight can generate new ideas and knowledge about the process of oral history, but also about the role and meaning of oral history in (post)modern society. Building upon this, the contemporary praxis of oral history (and its methodology and epistemology) could be enhanced in different ways, including the conscious reflection of contextual, situational, socio-historical and linguistic aspects of an oral historical interview.
Orální historie – ať už jako kvalitativní výzkumná metoda nebo disciplína sui generis – se v posledních desetiletích bezpochyby stává předmětem rostoucího zájmu akademiků, badatelů, ale také různých organizací a veřejnosti jako takové. Přibývá orálně historických projektů, které jsou realizovány v odlišných kontextech a regionech, zaměřují se na rozmanitá témata a historická období moderních dějin, vzájemně se liší rozpočtem, výzkumnými otázkami i technologickým zázemím. V tomto příspěvku přistupuji k problematice orální historie ze sociologického hlediska a pokusím se nastínit možné sociologické přístupy ke zkoumání tohoto sociálního fenoménu, které by bylo možno souhrnně označit souslovím sociologie orální historie. Jelikož toto odvětví sociologie zatím není formálně utvořeno (natož institucionálně zakotveno), má můj příspěvek spíše charakter programový a ukazuje různé teoretické a epistemologické cesty, jimiž by zkoumání na tomto poli mohlo v budoucnu směřovat.
[ "Jakub Mlynář" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Technology as a support means for the (re)use of narrative data in social sciences
Technologie pro podporu (sekundární) analýzy narativních dat ve společenských vědách
One of the consequences of the “digital turn” is a growing amount of linguistic narrative data available on-line. Large portion of this data is generated directly by the individual social actors. But there is also a second realm, which is a result – or sometimes a by-product – of qualitative research and oral history projects. In my talk, based on the case study of “sociology of oral history”, I intend to provide an insight into the ways of possible usage of oral history data in sociology, but also outline the crucial role of ICT in facilitating these efforts. Quite often, the results of process-oriented oral history projects (i.e. oral history projects without a specific research question) are published on-line in various digital environments. “Digital oral history” usually includes digital archives of interview collections with metadata, search tools, user work-space, Web 2.0 features etc. Due to this practice, large quantity of data is potentially available for reuse and secondary analysis by researchers, who did not participate in the initial data collection process. The analysis of this data could be supported by a wide variety of ICT tools. However, within the broad field of social sciences, different disciplines have very different requirements for exploitable data. I will use the data from USC Shoah Foundation’s Visual History Archive (http://vhaonline.usc.edu) in the context of my own research experience, in an attempt to illustrate this complexity.
Jedním z důsledků "digitálního obratu" je narůstající objem jazykových narativních dat dostupných on-line. Ve svém příspěvku jsem se zaměřil na případovou studii z oblasti "sociologie orální historie", konkrétně na různé způsoby, jak může sociologie využívat orálně historická data. Naznačil jsem také klíčovou roli ICT v usnadňování tohoto využití na příkladu Archivu vizuální historie USC Shoah Foundation.
[ "Jakub Mlynář" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
To pay or to get paid: Enriching a Valency Lexicon with Diatheses
Platit nebo dostat zaplaceno: Diateze ve valenčním slovníku
Valency lexicons typically describe only unmarked usages of verbs (the active form); however verbs prototypically enter different surface structures. In this paper, we focus on the so-called diatheses, i.e., the relations between different surface syntactic manifestations of verbs that are brought about by changes in the morphological category of voice, e.g., the passive diathesis. The change in voice of a verb is prototypically associated with shifts of some of its valency complementations in the surface structure. These shifts are implied by changes in morphemic forms of the involved valency complementations and are regular enough to be captured by syntactic rules. However, as diatheses are lexically conditioned, their applicability to an individual lexical unit of a verb is not predictable from its valency frame alone. In this work, we propose a representation of this linguistic phenomenon in a valency lexicon of Czech verbs, VALLEX, with the aim to enhance this lexicon with the information on individual types of Czech diatheses. In order to reduce the amount of necessary manual annotation, a semi-automatic method is developed. This method draws evidence from a large morphologically annotated corpus, relying on grammatical constraints on the applicability of individual types of diatheses.
Valenční slovníky typicky obsahují informace o nepříznakovém užití sloves (aktivní formu); nicméně slovesa typicky vstupují do různých povrchových struktur. Zde se soustřeďujeme na popis změn ve valenční struktuře, a to změn spojených s diatezemi.
[ "Anna Vernerová", "Václava Kettnerová", "Markéta Lopatková" ]
https://www.semanticscholar.org/paper/b036ff4c6abf7cbc43c06aa02b4310938d7bde9d/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Twitter Crowd Translation -- Design and Objectives
Twitter Crowd Translation -- Návrh a cíle
The paper describes the design and implementation of a system for human and machine translation of tweets. In these early experiments, we limit the system to follow some selected sources from Ukraine (the source language is primarily Ukrainian and Russian, sometimes English).
Příspěvek popisuje návrh a první implementaci systému pro lidský a strojový překlad tweetů. V prvních experimentech se omezujeme na překlad vybraných zdrojů z Ukrajiny (zdrojovým jazykem je primárně ukrajinština a ruština, občas angličtina). Zdrojové tweety jsou distribuovány registrovaným 'překladatelům', překlady je možné na webu ručně hodnotit a dostatečně dobré překlady odesíláme jako tweety do světa. Výhledově všechny tyto kroky bude provádět i stroj, zatím máme zapojen jen strojový překlad.
[ "Eduard Šubert", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/ff10cfd842619fbcf8fb003cb586f07e86a9a6ae/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
A corpus-based finite-state morphological toolkit for contemporary Arabic
Sada morfologických nástrojů pro současnou arabštinu založená na korpusu a konečných automatech
We develop an open-source large-scale finite-state morphological processing toolkit (AraComLex) for Modern StandardArabic (MSA) distributed under the GPLv3 license (http://aracomlex.sourceforge.net). The morphological transducer is based on a lexical database specifically constructed for this purpose. In contrast to previous resources, the database is tuned to MSA, eliminating lexical entries no longer attested in contemporary use. The database is built using a corpus of 1,089,111,204 word tokens, a pre-annotation tool, machine learning techniques and knowledge-based pattern matching to automatically acquire lexical knowledge. Our morphological transducer is evaluated and compared to LDC’s SAMA(StandardArabic Morphological Analyser). We also develop a finite-state morphological guesser as part of a methodology for extracting unknown word forms, lemmatizing them, and giving them a priority weight for inclusion in the lexicon.
Tento článek prezentuje sadu morfologických nástrojů pro současnou arabštinu založenou na korpusu a konečných automatech a poskytovanou pod GPLv3 licencí.
[ "Mohammed Attia", "Pavel Pecina", "Antonio Toral", "Josef Genabith" ]
https://www.semanticscholar.org/paper/b76b7e776c9467bcfae448c241618e1cb86e53b3/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Annotation of coreference in Prague Czech-English Dependency Treebank
Anotace koreference v Pražském česko-anglickém závislostním korpusu
The technical report presents the guidelines for manual annotation of nominal coreference in the Prague Czech-English Treebank.
Technická zprava shrnuje pravidla anotace koreference v Pražském česko-anglickém závislostním korpusu.
[ "Anna Nedoluzhko", "Jiří Mírovský", "Eva Fučíková", "Jiří Pergler" ]
https://www.semanticscholar.org/paper/311f279be8102364a6e511adde3f251beab329bd/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
RU-EVAL-2014: EVALUATING ANAPHORA AND COREFERENCE RESOLUTION FOR RUSSIAN
RU-EVAL-2014: Rozřešení anafory a koreference pro ruštinu
The paper reports on the recent forum RU-EVAL—a new initiative for evaluation of Russian NLP resources, methods and toolkits. The first two events were devoted to morphological and syntactic parsing correspondingly. The third event was devoted to anaphora and coreference resolution. Seven participating IT companies and academic institutions submitted their results for the anaphora resolution task and three of them presented the results of the coreference resolution task as well. The event was organized in order to estimate the state of the art for this NLP task in Russian and to compare various methods and principles implemented for Russian. We discuss the evaluation procedure. The anaphora and coreference tasks are specified in the present work. The phenomena taken into consideration are described. We also give a brief outlook of similar evaluation events whose experience we lay upon. In our work we formulate the training and Gold Standard corpora construction guidelines and present the measures used in evaluation.
V článku se popisují principy evaluační soutěže na automatické rozřešení koreferenčních a anaforických vztahů pro ruštinu.
[ "Anna Nedoluzhko", "Svetlana Toldova", "Anna Roitberg", "Alina Ladygina", "Maria D. Vasilyeva", "Ilja Azerkovich", "Matvej Kurzukov", "Anastasija Ivanova", "Julia Grishina" ]
https://www.semanticscholar.org/paper/2106ce34bac42c7ee76684128125bbc404508c36/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
"VCHERA NASOCHINYALSYA VOROH STROK": PRODUCTIVE CIRCUMFIXAL INTENSIFYING PATTERNS IN RUSSIAN
"VCHERA NASOCHINYALSYA VOROH STROK": Produktivní intenzifikační cirkumfixy v ruštině
The current paper addresses verbal circumfixal derivation patterns in modern Russian. The discussion is focused on a series of circumfixes which trigger the intensified usage of the basic verb (~'keep doingP too much'). Derivatives built up by adding a prefix and a reflexive -OR to an imper-fective verb are examined. Although each prefix adds specific shades of meaning to the verb, such patterns are, however, claimed to share common features at different levels of linguistic analysis, such as morphology, syntax, and semantics. Furthermore, such patterns are highly productive in modern language; once certain constraints are fulfilled, an intensified derivative can be formed from any imperfective verb. This fact, along with the patterns in question sharing certain common features, allow us to argue that they can be considered inflectional, rather than derivational.
Příspěvek uvádí sadu slovesných předpon, které spolu s reflexním morfémem mění význam slovesa, a to vždy stejným způsobem. Předpony tvoří posloupnost podle stupně intenzity, kterou pozměňují akci popisovanou daným slovesem. Představujeme proces intenzifikace slovesa, popisujeme sémantiku cirkumfixů, morfologické, syntaktické a sémantické omezení na jejich použití v ruštině.
[ "Anna Nedoluzhko", "Anna Schwarz (Khoroshkina)" ]
https://www.semanticscholar.org/paper/2b9ddff6b81879adad6f0f68617503dc11dbbfb0/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Parmesan: Meteor without Paraphrases with Paraphrased References
Parmesan: Meteor bez parafrází s parafrázovanými referenčními větami
This paper describes Parmesan, our submission to the 2014 Workshop on Statistical Machine Translation (WMT) metrics task for evaluation English-to-Czech translation. We show that the Czech Meteor Paraphrase tables are so noisy that they actually can harm the performance of the metric. However, they can be very useful after extensive filtering in targeted paraphrasing of Czech reference sentences prior to the evaluation. Parmesan first performs targeted paraphrasing of reference sentences, then it computes the Meteor score using only the exact match on~these new reference sentences. It shows significantly higher correlation with human judgment than Meteor on the WMT12 and WMT13 data.
Tento článek popisuje Parmesan, náš příspěvěk na Workshop on Statistical Machine Translation 2014. Ukazuje, že parafrázovací tabulky Meteoru pro češtinu obsahují tolik šumu, že jejich použití ve skutečnosti může poškodit výkon metriky. Nicméně po důkladní filtraci mohou být velmi užitečné v cíleném parafrázovní referenčních vět předcházejícím evaluaci. Parmesan nejprve provede cílené parafrázování referenčních vět a poté spočítá Meteor score s pouze přímou shodou na těchto nových referencích. Na datech z WMT12 a WMT13 ukazuje signifikantně vyšší shodu s lidským hodnocením než Meteor.
[ "Petra Barančíková" ]
https://www.semanticscholar.org/paper/a376b6aa64573c6341a182ff9c3ca5c3e91897ca/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Improving Evaluation of English-Czech MT through Paraphrasing
Zpřesňování hodnocení strojového překladu z angličtiny do češtiny pomocí parafrází
In this paper, we present a method of improving the accuracy of machine translation evaluation of Czech sentences. Given a reference sentence, our algorithm transforms it by targeted paraphrasing into a new synthetic reference sentence that is closer in wording to the machine translation output, but at the same time preserves the meaning of the original reference sentence. Grammatical correctness of~the new reference sentence is provided by applying Depfix on newly created paraphrases. Depfix is a system for post-editing English-to-Czech machine translation outputs. We adjusted it to fix the errors in paraphrased sentences. Due to a noisy source of our paraphrases, we experiment with adding word alignment. However, the alignment reduces the number of paraphrases found and the best results were achieved by~a~simple greedy method with only one-word paraphrases thanks to their intensive filtering. BLEU scores computed using these new reference sentences show significantly higher correlation with human judgment than scores computed on the original reference sentences.
Tato práce prezentuje možnost zpřesnění hodnocení strojového překladu českých vět. Náš algoritmus dostává na vstupu referenční větu a upravuje jí tak, aby byla bližší výstupu strojového překladu a současně si zachovala původní význam a gramatickou správnost. Tu zajišťuje systém Depfix, který byl původně navržen pro post-editaci strojových překladů z angličtiny do češtiny, zde upravený pro opravu chyb specifických pro parafrázovaní. Kvůli šumu ve zdrojích parafrázích experimentujeme se zarovnáváním slov. Ale jako nejlepší metoda se ukázala prostá hladová metoda s pouze jednoslovnými parafrázemi, díky jejich důkladnému filtrování. BLEU spočítané na nových referencích vykazuje významně vyšší korelaci s lidským hodnocením než skóre počítané na původních referencích.
[ "Petra Barančíková", "Rudolf Rosa", "Aleš Tamchyna" ]
https://www.semanticscholar.org/paper/ad9ff3ddc392ac6067fb6871afbadc2086b562fe/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Machine Translation within One Language as a Paraphrasing Technique
Strojový překlad v rámci jednoho jazyka jako technika pro parafrázování
We present a method for improving machine translation (MT) evaluation by targeted paraphrasing of reference sentences. For this purpose, we employ MT systems themselves and adapt them for translating within a single language. We describe this attempt on two types of MT systems -- phrase-based and rule-based. Initially, we experiment with the freely available SMT system Moses. We create translation models from two available sources of Czech paraphrases -- Czech WordNet and the Meteor Paraphrase tables. We extended Moses by a new feature that makes the translation targeted. However, the results of this method are inconclusive. In the view of errors appearing in the new paraphrased sentences, we propose another solution -- targeted paraphrasing using parts of a rule-based translation system included in the NLP framework Treex.
Představujeme metodu cíleného parafrazování pro zpřesnění hodnocení strojového překladu. V jejím rámci, využíváme strojový překld samotný a upravujeme jej tak, aby překládal v rámci jediného jazyka. Popisujeme tento přístup na dvou typech strojového překladu - statistickém a pravidlovém. V rámci statistického překladu experimentujeme s Mojžíšem, volně dostupným nástrojem pro statický strojový překlad. Překladové modely tvoříme uměle ze dvou dostupých zdrojů českých parafrází - českého WordNetu a parafrázovacích tabulek Meteor. Rozšiřujeme Mojžíše o nový rys, který vynucuje cílené parafrázování. Bohužel, naše výsledky jsou spíše negativní. S ohledem na chyby, které se objevily v parafrázovaných větách, navrhujeme nové řešení - parafrázování pomocí pravidlového systému Treex.
[ "Petra Barančíková", "Aleš Tamchyna" ]
https://www.semanticscholar.org/paper/221032a890c0b16eabab8059d5095854f66d09da/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
CUNI at the ShARe/CLEF eHealth Evaluation Lab 2014
CUNI v ShARe/CLEF eHealth Evaluation Lab 2014
This report describes the participation of the team of Charles University in Prague at the ShARe/CLEF eHealth Evaluation Lab in 2014.
Tato zpráva popisuje účast týmu Univerzity Karlovy v Praze na ShARe/CLEF eHealth Evaluation Lab in 2014.
[ "Shadi Saleh", "Pavel Pecina" ]
https://www.semanticscholar.org/paper/60083791a1207e4977b8590d29f866525c9ef93d/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2014
How does the automatic speech recognition work?
Jak funguje rozpoznávání řeči?
How does automatic speech recognition work? How are the studies at MFF UK? What is the Alex spoken dialogue system? Notes on machine learning and artificial intelligence
Jak funguje rozpoznávání řeči? Jaké je studium na Matfyzu? Co je Alex? (dialogový systém) Povídání o machine learning a artificial intelligence
[ "Ondřej Plátek" ]
https://www.semanticscholar.org/paper/a5331e52d28ae82aeab02b587c2282677a87540b/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Speech recognition using Kaldi
Rozpoznávání řeči pomocí Kaldi
The topic of this thesis is to implement efficient decoder for speech recognition training system ASR Kaldi (http://kaldi.sourceforge.net/). Kaldi is already deployed with decoders, but they are not convenient for dialogue systems. The main goal of this thesis to develop a real-time decoder for a dialogue system, which minimize latency and optimize speed. Methods used for speeding up the decoder are not limited to multi-threading decoding or usage of GPU cards for general computations. Part of this work is devoted to training an acoustic model and also testing it in the "Vystadial" dialogue system.
Tématem této práce je implementace výkonného rozpoznávače v open-source systému trénování ASR Kaldi (http://kaldi.sourceforge.net/) pro dialogové systémy. Kaldi již obsahuje ASR dekodéry, které však nejsou vhodné pro dialogové systémy. Hlavními důvody jsou jejich malá optimalizace na rychlost a jejich velké zpoždění v generování výsledku po ukončení promluvy. Cílem této práce je proto vyvinutí real-time rozpoznávače pro dialogové systémy optimalizovaného na rychlost a minimalizujícího zpoždění. Zrychlení může být realizováno například pomocí multi-vláknového dekó- dování nebo s využitím grafických karet pro obecné výpočty. Součástí práce je také příprava akustického modelu a testování ve vyvíjeném dialogovém systému ”Vystadial”.
[ "Ondřej Plátek" ]
https://www.semanticscholar.org/paper/e8e50c253a4a1c0e8509c122f4afa707706c1e4f/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Speech recognition using Kaldi
Rozpoznávání řeči pomocí Kaldi
We present an extension of the Kaldi automatic speech recognition toolkit to support on-line speech recognition. The resulting recogniser supports acoustic models trained using state-of-the-art acoustic modelling techniques. (MFCC, MLLT+LDA, BMMI) As the recogniser produces word posterior lattices, it is particularly useful in statistical dialogue systems, which try to exploit uncertainty in the recogniser's output. The evaluation is focused on on-line speech recognition in a dialogue system Alex for the "public Czech transportation" domain. The dialogue system is available at a public toll-free 800 899 998 line.
Představujeme rozšíření Kaldi automatického rozpoznávání řeči toolkit pro podporu on-line rozpoznávání řeči. Výsledný rozpoznáč podporuje state-of-the-art akustické modely. (MFCC, MLLT + LDA, BMMI) Vzhledem k tomu, že rozpoznávání produkuje slovní posteriorní lattice, je užitečné zejména v systémech statistických dialogu. V evaluaci se zaměříme na on-line rozpoznávání řeči v dialogovém systému Alex pro doménu "veřejné dopravy České republiky". Dialogový systém je k dispozici na veřejném bezplatném 800 899 998 řádek.
[ "Ondřej Plátek" ]
https://www.semanticscholar.org/paper/e8e50c253a4a1c0e8509c122f4afa707706c1e4f/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Free on-line speech recogniser based on Kaldi ASR toolkit producing word posterior lattices
Opensource on-line rozpoznávač řeči založený na Kaldi ASR toolkity dekódující slovní posteriorní svazy
This paper presents an extension of the Kaldi automatic speech recognition toolkit to support on-line recognition. The resulting recogniser supports acoustic models trained using state-of-the-art acoustic modelling techniques. As the recogniser produces word posterior lattices, it is particularly useful in statistical dialogue systems, which try to exploit uncertainty in the recognizer's output. Our experiments show that the on- line recogniser performs significantly better in terms of latency when compared to a cloud-based recogniser.
V tomto článku je presentováno rozšíření Kaldi toolkitu pro rozpoznávání řeči, které podporuje on-line rozpoznávání. Vysledný rozpoznávač řeči podporuje moderní modelovací metody. Jelikož rozpoznávač vytváří slovní posteriorní svazy, je obzvlášt vhodný pro statistické dialogové systémy, které umí pracovat s nejistotou rozpoznávače řeči. Experimenty ukazují, že online rozpoznávač má výrazně lepší výsledky vzhledem k latenci pri srovnání s rozpoznávačem v cloudu.
[ "Ondřej Plátek", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/1126af98ea55dc4dcbf0f9a5a83abe915c80ad97/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Integration of an On-line Kaldi Speech Recogniser to the Alex Dialogue Systems Framework
Integrace on-line Kaldi rozpoznávače řeči do frameworku dialogových systémů Alex
This paper describes the integration of an on-line Kaldi speech recogniser into the Alex Dialogue Systems Framework (ADSF). As the Kaldi OnlineLatgenRecogniser is written in C++, we first developed a Python wrapper for the recogniser so that the ADSF, written in Python, could interface with it. Training scripts for acoustic and language modelling were developed and integrated into ADSF, and acoustic and language models were build. Finally, optimal recogniser parameters were determined and evaluated. The dialogue system Alex with the new speech recogniser is evaluated on Public Transport Information (PTI) domain.
Tento článek popisuje integrace on-line Kaldi rozpoznávače řeči do frameworku dialogovýho systémů Alex (ADSF). Jelikož Kaldi OnlineLatgenRecogniser je implementován v C++, bylo nejprve nutno převést API rozpoznávače do jazyka Python. Skripty pro trénování akustických a jazykových modelů byly integrovány do ADSF a příslušné modely byly natrénovány. Na závěr optimální parametry rozpoznávače byly zvoleny z dat a ověřeny na "Public Transport Information (PTI)" doméně.
[ "Ondřej Plátek", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/14a6c0a97c82c8d502ce53a60698ce630bafb294/
null
null
null
null
null
null
null
null
null
null
null
null
en
2014
Deep and Phrase-Based MT into Czech
Hloubkový a frázový strojový překlad do češtiny
An overview of problems caused in MT by rich target-side morphology, and our techniques of addressing them.
Přehled problémů, které ve strojovém překladu způsobuje bohatá morfologie na cílové straně a naše techniky, jimiž se s nimi vyrovnáváme.
[ "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/aeaf5726ce467b77f46fddbd24825f5ad834b49f/
null
null
null
null
null
null
null
null
null
null
null
null