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
Towards comprehensive approaches to discourse-structuring devices: PDiT - GECCo Scientific mission
K ucelenému popisu diskurzních prostředků: PDiT - GECCo vědecká mise
In the presentation, the two approaches to discourse-structuring devices have been compared and analysed: PDiT and GECCo
V této prezentaci jsme představili první výsledky srovnání dvou anotačních schémat, PDiT a GECCo v oblasti zpracování diskurzních konektorů.
[ "Anna Nedoluzhko", "Ekaterina Lapshinova-Koltunski", "Kerstin Anna Kunz" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Coreference chains in Czech, English and Russian: Preliminary findings
Koreferenční řetězce v češtině, angličtině a ruštině: Předběžná zjištění
Tento článek je pilotní srovnavací výzkum koreferenčních řetězců v češtině, angličtině a ruštině. Podrobili jsme analýze 16 srovnatelných textů ve třech jazycích. Naší motivací bylo zjistit lingvistickou strukturu koreferenčních řetězců v těchto jazycích a určit, které faktory ovlivňují tuto strukturu.
This paper is a pilot comparative study on coreference chaining in three languages, namely, Czech, English and Russian. We have analyzed 16 parallel English-Czech newspaper texts and 16 texts in Russian (similar to the English-Czech ones in length and topics). Our motivation was to find out what the linguistic structure of coreference chains in different languages is and what types of distinctions we should take into account for advancing the development of systems for coreference resolution. Taking into account theoretical approaches to the phenomenon of coreference we based our research on the following assumption: the recognition of coreference links for different structural types of noun phrases is regulated by different language mechanisms. The other starting point was that different languages allow pronominal chaining of different length and that coreference chains properties differ for the languages with different strategies for zero anaphora and different systems for definiteness marking. This work reports our first findings within the task of the structural NP types’ distribution comparison in three languages under analysis.
[ "Anna Nedoluzhko", "Svetlana Toldova", "Michal Novák" ]
https://www.semanticscholar.org/paper/c84ba480831ee3d5b0963110cb2afb46911a22eb/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Slovakoczech NLP workshop (SloNLP 2015)
Slovenskočeský NLP workshop (SloNLP 2015)
SloNLP is a workshop focused on Natural Language Processing (NLP) and Computational Linguistics. Its primary aim is to promote cooperation among NLP researchers in Slovakia and Czech Republic. The topics of the workshop include automatic speech recognition, automatic natural language analysis and generation (morphology, syntax, semantics, etc.), dialogue systems, machine translation, information retrieval, practical applications of NLP technologies, and other topics of computational linguistics.
SloNLP je workshop speciálně zaměřený na zpracování přirozeného jazyka a počítačovou lingvistiku. Jeho hlavním cílem je podpořit spolupráci mezi výzkumníky v oblasti NLP v Česku a na Slovensku. Mezi témata workshopu patří automatické rozpoznávání mluvené řeči (ASR), automatická analýza a generování přirozeného jazyka (morfologie, syntax, sémantika...), dialogové systémy (DS), strojový překlad (MT), vyhledávání informací (IR), praktické aplikace NLP technologií, a další témata počítačové lingvistiky.
[ "Petra Barančíková", "Rudolf Rosa" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Targeted Paraphrasing on Deep Syntactic Layer for MT Evaluation
Cílené parafrázování na rovině hloubkové syntaxe pro evaluaci strojového překladu
In this paper, we present a method of improving quality of machine translation (MT) evaluation of Czech sentences via targeted paraphrasing of reference sentences on a deep syntactic layer. For this purpose, we employ NLP framework Treex and extend it with modules for targeted paraphrasing and word order changes. Automatic scores computed using these paraphrased reference sentences show higher correlation with human judgment than scores computed on the original reference sentences.
Představujeme vylepšení evaluace strojového překladu do češtiny pomocí cíleného parafrázování referenčních vět na rovině hloubkové syntaxe.
[ "Petra Barančíková", "Rudolf Rosa" ]
https://www.semanticscholar.org/paper/8dce90534a1cbc997960a0b4a76d3b348be2bbf6/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Cross language information retrieval systems
Systémy pro získávání informací napříč jazyky
In this paper we will explore different approaches used in cross languages information retrieval (CLIR) systems, mainly dictionary-based and corpus based systems. Then we will focus on CLIR systems which use statistical machine translation (SMT) systems to translate queries into collection language. Different approaches which use SMT are studied in this paper including the using of SMT as black box and looking inside the box of the SMT system to tune it to get better performance in IR’s point of views rather than using SMT system to give better results for human’s point of views.
Článek se zabývá různými přístupy ke získávání informací napříč jazyky, zejména slovníkovými a korpusovými přístupy. Zaměříme se také na systémy, které používají statistický strojový překlad pro překládání dotazů nad kolekcí jazyků.
[ "Shadi Saleh" ]
https://www.semanticscholar.org/paper/372fd06ec2b25acf0f017db69c0ae4cc545a787c/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
CUNI at the CLEF eHealth 2015 Task 2
CUNI v CLEF eHealth 2015 Task 2
This report describes the participation of the team of Charles University in Prague at the CLEF eHealth 2015 Task 2.
Tato zpráva popisuje účast týmu Univerzity Karlovy v Praze na CLEF eHealth 2015 Task 2.
[ "Shadi Saleh", "Feraena Bibyna", "Pavel Pecina" ]
https://www.semanticscholar.org/paper/f756f5cef75ea1a66912ec4d2aa2b83a938f75d4/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
PechaKucha night Ústí nad Labem - How to teach machines to talk?
PechaKucha Ústí nad Labem - Jak počítače naučit mluvit
Introducing the task of speech recognition, problems related with spoken dialogue systems. Presentation of Alex spoken dialogue system.
Představení problému rozpoznávání řeči a dialogových systémů. Prezentace dialogového systému Alex.
[ "Ondřej Plátek" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Self awareness for better common ground
Sebeuvědomění pro lepší vzájemné porozumění
Having set up several dialogue systems for multiple domains we realized that we need to introduce each system thoroughly to the users in order to avoid misunderstanding. Even with well-described systems, users still tend to request out-of-domain information and are often confused by the system response. In our ongoing work, we try to address these issues by allowing our conversational agent to speak about its abilities. Our agent is able to simply describe what it understands and why it decided to perform one of its summary actions. In this short paper, we present our current system architecture.
Dialogovový agent musí být uživateli velmi důkladně představen, aby se zabránilo četným nedorozumněním. Přesto však se uživatelé často ptají na funkcionalitu, které systém nerozumí a neumí jí poskytnout. V naší pokračující práci adresujeme tento problém tak, že agent umí popsat své možnosti a zároveň vysvětluje proč zvolil své akce.
[ "Ondřej Plátek", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/26677e57ff85481d7b0e4e2a79d64381b88d9a3f/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Self-awareness: Towards Extracting Knowledge from Dialogue
Extrakce znalostí z dialogu
In our ongoing work, we try to build an conversational agent which is able to chat about its abilities. Our implemented agent is able to simply describe what it understands and why it decided to perform one of its summary actions. We focus on self-awareness of our agent because it helps defining the agent’s domain, and its architecture to its users. In this paper we introduce our current implementation, outline our goals and describe our future intentions.
Článek popisuje probíhající výzkum zaměřený na vývoj konverzačního agenta, který je schopen diskutovat o svých schopnostech. Náš implementovaný agent je schopen jednoduše popsat, čemu rozumí a proč se rozhodl pro kterou akci. Zaměřujeme se na sebeznalost agenta, protože to pomůže popsat uživatelům agentovu doménu.
[ "Ondřej Plátek", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/7545ea419588a9b7c7653d7f507ee67444d5c149/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
"I was born in Czechoslovakia..." Twentieth century in the memories of witnesses from the Carpathian Ruthenia saved in the USC Shoah Foundation's archive
"Narodila jsem se v Československu..." Dvacáté století ve vzpomínkách pamětnic z Podkarpatské Rusi uložených v archivu USC Shoah Foundation
The presentation provided an overview of USC Shoah Foundation's Visual History Archive and related educational activities and projects. Two short videoclips from three witnesses' life story narratives captured different aspects of the history of Carpathian Ruthenia as a former part of Czechoslovakia.
Prezentace představila Archiv vizuální historie USC Shoah Foundation a související vzdělávací aktivity a projekty. Dva krátké sestřihy z úryvků vzpomínek tří pamětnic přiblížily z různých úhlů dějiny Podkarpatské Rusi jako někdejší součásti Československa.
[ "Martin Šmok", "Jakub Mlynář" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
How machine translation works
Jak pracuje automatický překlad
An introduction to machine translation for employees of governmental and other public institutions.
Představení technik strojového překladu pro pracovníky státní správy a veřejných institucí.
[ "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/fc94b4111debe8a51c8350399ffea17486b7b20c/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Machine translation
Strojový překlad
The chapter on Machine translation in Oxford Handbook of Inflection starts by describing the few key approaches to machine translation and proceeds over individual stages of the process, highlighting the problems caused by the richness of word forms.
Kapitola v příručce o tvarosloví Oxford Handbook of Inflection se věnuje problematice bohatosti slovních forem v úloze strojového překladu. Po úvodu popisujícím několik klasických typů strojového překladu se podrobně zabývá jednotlivými fázemi či kroky strojového překladu a potížím, které v nich pestrost tvarosloví přináší.
[ "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/1518039b5001f1836565215eb047526b3ac7f462/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Findings of the 2015 Workshop on Statistical Machine Translation
Výsledky Workshopu o statistickém strojovém překladu v roce 2015
This paper presents the results of the WMT15 shared tasks, which included a standard news translation task, a metrics task, a tuning task, a task for run-time estimation of machine translation quality, and an automatic post-editing task. This year, 68 machine translation systems from 24 institutions were submitted to the ten translation directions in the standard translation task. An additional 7 anonymized systems were included, and were then evaluated both automatically and manually. The quality estimation task had three subtasks, with a total of 10 teams, submitting 34 entries. The pilot automatic postediting task had a total of 4 teams, submitting 7 entries.
Článek představuje výsledky společných úloh WMT15 -- překladu novinových textů, odhadu kvality překladu a automatické posteditace. Do standardní překladové úlohy v 10 překladových směrech se letos zapojilo 68 systémů strojového překladu z 24 institucí. Zároveň bylo vyhodnoceno 7 anonymizovaných systémů. Úloha odhadu kvality překladu měla 3 podúlohy, kterých se zúčastnilo 10 týmů a celkem 34 systémů. Pilotní úloha automatické posteditace měla 4 týmy, které odeslaly 7 systémů
[ "Ondřej Bojar", "Rajen Chatterjee", "Christian Federmann", "Barry Haddow", "Matthias Huck", "Chris Hokamp", "Philipp Koehn", "Varvara Logacheva", "Christof Monz", "Matteo Negri", "Matt Post", "Carolina Scarton", "Lucia Specia", "Marco Turchi" ]
https://www.semanticscholar.org/paper/feb420a4ac7c5719d51480053cd3e8669d5f2062/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
MT for Industry from Researchers’ Point of View
Strojový překlad pro průmysl z pohledu výzkumníka
The talk presented solutions for successful transfer of machine translation technology from research to commercial use and its evaluation in the context of translation and localization of documentation in the area of information technology.
Příspěvek představil problematiku strojového překladu při přenosu technologie z výzkumné sféry do rutinního komerčního využití a způsob hodnocení takového překladu v kontextu lokalizace dokumentace v oblasti ifnormačních technologií.
[ "Ondřej Bojar", "Jan Hajič" ]
https://www.semanticscholar.org/paper/5566726e5b83e868e529338a5dd10d33eae03c18/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
CUNI in WMT15: Chimera Strikes Again
CUNI ve WMT15: Chiméra opět zasahuje
This paper describes our WMT15 system submission for the translation task, a hybrid system for English-to-Czech translation. We repeat the successful setup from the previous two years.
Článek popisuje náš systém pro Translation Task WMT15. Jedná se o hybridní systém pro anglicko-český překlad. Využíváme úspěšnou konfiguraci z předchozích dvou let.
[ "Ondřej Bojar", "Aleš Tamchyna" ]
https://www.semanticscholar.org/paper/b95c8e996b37d3dc81e29e44b2adde23bfb4d951/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
TMODS:ENG-CZE - System for Translation of Text for the English-Czech Language Pair
TMODS:ENG-CZE - systém pro překlad souvislého textu pro jazykový pár en-cs
Tools and models for two machine translation systems: English->Czech translation (mainly longer text) and Czech->English translation (short queries as well as text). In the first case, the models are the statistical component of the Chimera system. Models were pruned and stored in a format which allows for very fast translation, at the expense of possible slight reduction of output quality.
Nástroje a modely pro dva překladové systémy: anglicko-český překlad (především souvislých textů) a česko-anglický překlad (krátkých dotazů i textů). V prvním případě jde o statistickou komponentu překladového systému Chiméra. Modely byly zmenšeny a uloženy ve formátu, který umožňuje velmi rychlý překlad, za cenu možné drobné ztráty kvality výstupu.
[ "Ondřej Bojar", "Aleš Tamchyna" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
A Summary of Research Activities: Technologies – Demands – Gaps – Roadmaps
Přehled výzkumných aktivit v EU: technologie, požadavky, mezery, cestovní mapy
A summary of research activities in the area of machine translation in the EU (Technologies – Demands – Gaps – Roadmaps) has been presented, including contributions from multiple other EU-funded projects.
Shrnutí aktivit v oblasti strojového překladu v EU včetně příspěvků několika projektů v této oblasti podporovaných Evropskou komisí, a směřování této oblasti směrem k jednotnému digitálnímu trhu v EU.
[ "Jan Hajič" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Abstract Meaning Representation across Languages
Abstraktní reprezentace významu pro více jazyků
Abstract Meaning Representation is a newly developed formalism for representing meaning, which abstracts from syntax and some other phenomena but is (still) language-dependent. It has been developed by a consortium of mostly U.S. universities, with team members including Martha Palmer, Kevin Knight, Philipp Koehn, Ulf Hermjakob, Kathy McKeown, Nianwen Xue and others. In the talk, the basic facts about the AMR will be presented, and then comparison will be made for Czech and English as carried out in detail on a small 100-sentence corpus; some examples from Chinese-English comparison will also be shown. In addition, AMR will be compared to the deep syntactic representation used in the set of Prague Dependency Treebanks (again, for Czech And English), and observations will be made about the level of abstraction used in these two formalisms. Plans for future studies and possible corpus annotation work in the nearest future will also be mentioned. The work reported has been done primarily by the CLAMR (Cross-Lingual AMR) team led by Martha Palmer of UCB at the Johns Hopkins Summer workshop in 2014.
Abstraktní reprezentace významu, vyvinutá konsorciem amerických univerzit, slibuje vyšší úroveň reprezentace významu a jeho přiblížení i mezi vzdálenými jazyky. Přednáška představila analýzu různých problémů, které přitom vznikají.
[ "Jan Hajič" ]
https://www.semanticscholar.org/paper/37fd65d01fbbde65c02800864c2f32770c30d17f/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Abstract Meaning Representation: comparing languages and their representations
Abstraktní reprezentace významu: porovnání jazyků a jejich reprezentací
Abstract Meaning Representation is a newly developed formalism for representing meaning, which abstracts from syntax and some other phenomena but is (still) language-dependent. It has been developed by a consortium of mostly U.S. universities, with team members including Martha Palmer, Kevin Knight, Philipp Koehn, Ulf Hermjakob, Kathy McKeown, Nianwen Xue and others. In the talk, the basic facts about the AMR will be presented, and then comparison will be made for Czech and English as carried out in detail on a small 100-sentence corpus; some examples from Chinese-English comparison will also be shown. In addition, AMR will be compared to the deep syntactic representation used in the set of Prague Dependency Treebanks, and observations will be made about the level of abstraction used in these two formalisms. There is an ongoing work on possible conversion between the Prague deep syntactic representation and the AMR representation, and the main issues of such a conversion will also be described, as well as plans for future studies and possible corpus annotation work in the nearest future.
Abstraktní reprezentace významu se snazží podchytit formálně význam vět v přirozeném jazyce. V příspěvku se porovnává s existujícími reprezentacemi a rovněž i napříč jazyky. Hlavní porovnání vylo provedeno s formalismem Pražského závislostního korpusu.
[ "Jan Hajič" ]
https://www.semanticscholar.org/paper/408c37d4a6c6285df31b82ea9f5ae653ce5be04b/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Deep Linguistic Information in Machine Translation
Hloubková jazyková informace ve strojovém překladu
Fundamentals of Machine Translation technology using Deep language analysis using the traditional "Vauquois triangle" have been presented, which now use advanced statistical and machine learning techniques. In addition, latest WMT 2015 Shared Task competition results for the en-cs pair have been shown in which the Chimera system created at UFAL MFF UK won.
Byly prezentovány základy strojového překladu s použitím hloubkové jazykové analýzy podle tradičního systému "Vauquois trojúhelník", s využitím moderních statistických metod a metod strojového učení. Rovněž byly ukázány poslední výsledky, kdy hybridní systém Chiméra vytvořený na ÚFAL MFF UK v rámci různých projektů zvítězil v soutěži en-cs překladačá na WMT 2015 Shared Task.
[ "Jan Hajič" ]
https://www.semanticscholar.org/paper/b218cc96f3e78dba9a5b7746e95d651db97a1dfd/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Evaluation of Research Programs at the Technology Agency of the Czech Republic
Hodnocení výzkumných programů v Technologické agentuře České Republiky
Technology Agency of the Czech Republic is less than six-years-old funding agency charged with promoting applied research and innovation. Its responsibility is to create programs, to be approved by the government, and then execute the whole lifecycle of each program -- from calls, ex-ante evaluation or proposals, handling the awards organizationally and financially throughout projects' execution, and performing ex-post evaluation of projects and the whole programs. While the ex-ante evaluation system has been described at AEA 2014, this year we plan to present results of a mid-term evaluation of a program for applied research in the area of social sciences and of the top program of the agency, the Centers of Competence program.
Technologická agentura České republiky již šest let poskytuje podporu v oblasti aplikovaného výzkumu. Je odpovědná za tvorbu programů v oblasti aplikovaného výzkumu. V prezentaci je představeno hodnocení programu Center kompetence TAČR, která byla hodnocena v polovině jejich existence.
[ "Jan Hajič" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Experience of the Institute of Formal and Applied Linguistics: Framework ICT projects
Zkušenosti Ústavu formální a aplikované lingvistiku MFF UK v Praze s rámcovými programu EU v oblasti ICT
EU Framework and Horizon 2020 projects carried out in the Institute of Formal and Applied Linguistics of the Faculty of Mathematics and Physics, Charles University in Prague, have been presented.
V příspěvku byly prezentovány zušenosti Ústavu formáln a aplikované lingvistiky MFF UK s projekty v rámcových programech EU a zejména v programu Horizon 2020.
[ "Jan Hajič" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Language technology for the medical domain
Jazykové technologie pro oblast medicíny a lékařství
The talk presented activities in the area of language technology and specifically machine translation in the medical domain in the context of EC-funded projects.
Přednáška představila projekty a další aktivity v oblasti jazykových technologií a speciálně strojového překladu pro tématickou oblast medicíny a lékařství.
[ "Jan Hajič" ]
https://www.semanticscholar.org/paper/dc6260851efb43d87c9030d42cd5a68936961c39/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
LINDAT/CLARIN: data and technology for research based on written and spoken language analysis
LINDAT/CLARIN: data a technologie pro výzkum založený na analýze psaného a mluveného jazyka
The project LINDAT/CLARIN (Building and operation of a Czech node of the pan-European infrastructure for research) is a Czech node of the international network Clarin ERIC (Common Language Resources and Technology Infrastructure), headquartered in Utrecht, Netherlands. The goal of Clarin is to provide open access to research data, technology and services.
Projekt výzkumné infrastruktury LINDAT/CLARIN (“Vybudování a provoz českého uzlu panevropské infrastruktury pro výzkum”) je koncipován jako český „uzel“ mezinárodní sítě Clarin (Common Language Resources and Technology Infrastructure), která je zřízena ve formě ERIC (Evropské konsorcium pro výzkumnou infrastrukturu) se sídlem v Utrechtu v Holandsku. Cílem Clarin ERIC je otevřený přístup k jazykovým datům a technologiím zejména pro humanitní a společenské obory. LINDAT/CLARIN je rovněž kompatibilní s technologickou sítí META-SHARE (http://www.meta-net.eu), která je zaměřená na jazykové technologie a aplikace, čímž tyto dvě oblasti propojuje. LINDAT/CLARIN shromažďuje, zpracovává, anotuje (manuálně a automaticky) a uchovává jazyková data v českém jazykovém prostředí a poskytuje softwarové nástroje pro vyhledávání, analýzu a syntézu přirozeného jazyka. Použitá technologie a její kvalita a rozšiřitelnost je přímo aplikovatelná jak v humanitních a společenských vědách a výzkumu (jazykověda a interdisciplinární výzkum s jazykovou složkou, jako například formální a počítačová lingvistika, translatologie, lexikografie, sociolingvistika, psychologie, sociologie, historie, literární vědy, neurolingvistika, kognitivní vědy, umělá inteligence a strojové učení), tak ve výzkumu a vývoji jazykových technologií založených na statistických metodách (zpracování přirozeného jazyka, rozpoznávání a syntéza řeči a kombinovaná analýza obrazu, textu a multimédií obecně, vč. text and data mining“, strojového překladu a extrakce informace).
[ "Jan Hajič" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Machine Translation of Natural Languages
Strojový překlad přirozených jazyků
Machine Translation is as old as the field of Computational Linguistics itself. It is also a problem that has been predicted to „be solved in the next five years“ many times, but in fact it is not yet solved today. Machine translation has been naively considered a simple problem solvable by simple statistical means, then studied in depth by complicated but unsuccessful detailed sets of rules trying to describe all details of natural language use, only to return to statistical approach on a completely different level, using a combination of linguistic analysis and powerful machine learning algorithms.
Strojový překlad přirozených jazyků je problém stejně starý jako počítačová lingvistika samotná, a v jejím rámci je to problém, který byl mnohokrát podceněn jak z hlediska metodologického, tak z hlediska doby potřebné k jeho vyřešení – a ani dnes vyřešen není. Pohled na překlad prošel vývojovou spirálou od primitivních pokusů o jednoduché statistické řešení přes nesmírně komplikované algoritmy popisující všechny detaily a výjimky v přirozeném jazyce až k dnešním přístupům, které kombinují statistický a lingvistický přístup.
[ "Jan Hajič" ]
https://www.semanticscholar.org/paper/bc2c7f19877344670cfd99f1ed66849050be3b40/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
The LINDAT/CLARIN large infrastructure: data and services
Výzkumná infrastruktura LINDAT/CLARIN: data a služby
LINDAT/CLARIN, as a node of the pan-European research infrastructure Clarin ERIC, has been presented. Its repository has been featured together with data archivation techniques and related web services and applications.
V přednášce byl prezentován LINDAT/CLARIN jako uzel mezinárodní infrastrukturní sítě Clarin ERIC. Zejména byl prezentován repozitář LINDAT/CLARIN a jeho možnosti ukládání a archivace jazykových a jiných dat, a s ním spojené služby.
[ "Jan Hajič" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Treebanks and MWEs
Treebanky a víceslovné výrazy
Presented were existing syntactically analyzed treebanks, such as the Penn Treebank and the Prague family of dependency Treebanks, and their relation to Multiword Expressions.
V přednášce byly prezentovány známé a používané korpusy se zachycenou syntaktickou analýzou (Penn Treebank, Pražské závislostní korpusy) a jejich vztah k analýze váceslovných výrazů.
[ "Jan Hajič" ]
https://www.semanticscholar.org/paper/12397af7a96264b8d6697dee7b0a61d24d06190b/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
A-GOOGLE: annotation software and data
A-GOOGLE: software a data pro anotaci
Licence agreement between Charles University in Prague and Google Inc., effective Dec. 2, 2014, duration unless terminated, PO No. 270122, invoiced Jan. 2, 2015. The license covers the right to use the data of PDT 2.5 and the associated software "MorphoDiTa" for morphological tagging, for commercial purposes. The same data and software is otherwise available for free through the LINDAT/CLARIN repository for non-commercial use under the appropriate variant of the Creative COmmons license.
Licence mezi UK v Praze a Google Inc., platná od 2. 12. 2014 na dobu neurčitou, PO 270122, fakturováno 2. 1. 2015. Obsahuje právo na užití dat PDT 2.5 a asociovaného software "MorphoDiTa" pro morfologické značkování na základě PDT 2.5, a to pro komerční účely. Jedná se o identické verze uložené v repozitáři LINDAT/CLARIN s nekomerční licencí typu Creative Commons.
[ "Jan Hajič", "Ondřej Bojar", "Markéta Lopatková" ]
https://www.semanticscholar.org/paper/a765b136bae55f9120805418e50dc08deceb5e80/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Deletions and node reconstructions in a dependency-based mutlilevel annotation scheme
Elipsy a rekonstruované uzly ve víceúrovňovém závislostním anotačním schématu
The aim of the present contribution is to put under scrutiny the ways in which the so-called deletions of elements in the surface shape of the sentence are treated in syntactically annotated corpora and to attempt at a categorization of deletions within a multilevel annotation scheme.
Cílem tohoto příspěvku je podrobně prozkoumat, jak se v syntakticky anotovaných korpusech zachází s prvky věty, které jsou na povrchu vypuštěné, a pokusit se o kategorizaci výpustek v rámci víceúrovňového anotačního schématu.
[ "Jan Hajič", "Eva Hajičová", "Marie Mikulová", "Jiří Mírovský", "Jarmila Panevová", "Daniel Zeman" ]
https://www.semanticscholar.org/paper/b36d39ea1b54117e0ec4b4cc94f2e8a00dfe3673/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
Translation of the VHA Archive: project AMALACH
Překlad archívu VHA:projekt AMALACH
Project Amalach has resulted in software for cross=language search in the VHA archive. It also resulted in the Czech version of the VHA thesaurus. The talk has presented the topic of audio search in the VHA archive.
Projekt Amalach vytvořil software pro prohledávání nahrávek archivu VHA v češtině a angličtině. Součástí projektu byl rovněž překlad tezauru použitého pro indexaci archivu. Přednáška uvedla celou problematiku vyhledávání a audioarchivu VHA.
[ "Jan Hajič", "Aleš Tamchyna", "Pavel Ircing", "Josef Psutka" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
Prof. Dr. Fr. Daneš has passed away (23.7.1919-18.3.2015)
Odešel prof. dr. Fr. Daneš (23.7.1919-18.3.2015)
Obituary of prof. dr. František Daneš,DrSc. A survay of his wide scale of linguistic interests and of his important contributions to the Czech and word linguistics.
Nekrolog významného českého lingvisty prof. dr. Františka Daneše, DrSc. Zamyšlení nad rozpětím jeho zájmů a nad jeho přínosy české a světové lingvistice.
[ "Jarmila Panevová" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
Adverbial of Comparison in Czech
Příslovečné určení srovnání v češtině
Czech adverbial constructions introduced by the expressions jako and než are understood as grmmaticalized ellipsis. Their undelying structures are proposed in the shape of reconstructed constructions containing either the repeated items from the governing predication or by the new artfitial nodes with general semantics.
Srovnávací struktury uvozené v češtině výrazy jako a než se reprezentují jako gramatikalizované elipsy, jejichž zdrojem jsou rekonstruované predikační struktury s opakováním slov z řídící predikace nebo zastoupené umělým uzlem s obecnou sémantickou platností. Dále se analyzuje srovnání uvozeneé sekundárními předložkami.
[ "Jarmila Panevová", "Marie Mikulová" ]
https://www.semanticscholar.org/paper/8d3415a32f4d8b65ac10c3e83cc42b1e25f04528/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Reconstruction of Deletions in a Dependency-based Description of Czech: Selected Issues
Rekonstrukce elips v závislostním popisu češtiny: Vybrané příklady
In this contribution the treatment of selected types of deletions in Czech are presented. The pro-dropped constructions,absent subjects in controlled infinitive constructions, selected types of Czech "small clauses" and comparison constructions are analyzed and exemplified.
V příspěvku se prezentuje způsob zachycení vybraných typů elips v češtině v rámci FGP. Jde zejména o vypouštění zájmenného subjektu v 1. a 2. osobě, o nepřítomnost subjektu v kontrolovaných infinitivních konstrukcích,o konstrukce uvozené "kromě" a "místo" a o srovnávací konstrukce.
[ "Eva Hajičová", "Marie Mikulová", "Jarmila Panevová" ]
https://www.semanticscholar.org/paper/882a44e9b35bbc6c01727a1eba3757f5b79298a4/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Prague Trees
Pražské stromy
Presentation covers PML format and tools using it (Treex, PML-TQ, TrEd). Using real annotated data as examples representation of constructions and searching for them using the tools was demonstrated.
Prezentace pokrývá formát PML (Prague Markup Language) pro reprezentaci liongvistických dat, nástroje pracující s tímto formátem (Treex, TrEd, PML-TQ) a na příkladech anotovaných dat ukazuje využití PML a nástrojů pro reprezentaci a vyhledávání lingvistických jevů.
[ "Pavel Straňák" ]
https://www.semanticscholar.org/paper/a279469ef9b04ae23a0df303a2d9947851ac575d/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Domain adaptation of statistical machine translation with domain-focused web crawling
Doménová adaptace statistického strojového překladu s doménově specifickým stahováním data z webu
In this paper, we tackle the problem of domain adaptation of statistical machine translation (SMT) by exploiting domain-specific data acquired by domain-focused crawling of text from the World Wide Web. We design and empirically evaluate a procedure for auto- matic acquisition of monolingual and parallel text and their exploitation for system training, tuning, and testing in a phrase-based SMT framework. We present a strategy for using such resources depending on their availability and quantity supported by results of a large-scale evaluation carried out for the domains of environment and labour legislation, two language pairs (English–French and English–Greek) and in both directions: into and from English. In general, MT systems trained and tuned on a general domain perform poorly on specific domains and we show that such systems can be adapted successfully by retuning model parameters using small amounts of parallel in-domain data, and may be further improved by using additional monolingual and parallel training data for adaptation of language and translation models. The average observed improvement in BLEU achieved is substantial at 15.30 points absolute.
V tomto článku se zabýváme problémem doménové adaptace statistického strojového překladu (SMT) s využitím doménově specifických dat získaných cíleným prohledáváním (crawling) v Internetu. Navrhujeme a empiricky vyhodnocujeme proces automatického získávání jednojazyčných a paralelních textů, stejně jako jejich využití v trénování, optimalizaci i testování v rámci frázového SMT. Navrhujeme strategii, jak tyto zdroje využít v závislosti na jejich dostupnosti a množství, která ja podpořena výsledky rozsáhlé evaluace. Ta byla provedena v doménách legislativy pro životní prostředí a práci, ve dvou jazykových párech (angličtina-francouzština a angličtina-řečtina) a v obou směrem: z angličtiny i do angličtiny. Obecně lze tvrdit, že systémy SMT trénované a optimalizované na datech z obecné domény na specifických doménách dosahují špatných výsledků. Ukazujeme, že takové systémy lze úspěšně adaptovat optimalizací parametrů modelu s využitím malého množství paralelních dat z cílové domény, a dále je lze zlepšit přidáním jednojazyčných i paralelních dat pro adaptaci jazykového, resp. překladového modelu. Průměrné pozorované zlepšení v BLEU o 15,3 absolutních bodů je významné.
[ "Pavel Pecina", "Antonio Toral", "Vassilis Papavassiliou", "Prokopis Prokopidis", "Aleš Tamchyna", "Andy Way", "Josef Genabith" ]
https://www.semanticscholar.org/paper/5d6e74a247fe8d90f1a3f84643aeae17eb5ae79a/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
KWEC Keyword extractor
KWEC Extractor klíčových slov
A webservice supporting indexation of book table of contents. For the National Technical Library.
Webová služba pro podporu indexování naskenovaných obsahů knih. Pro Národní technickou knihovnu.
[ "Pavel Pecina", "Jindřich Libovický" ]
https://www.semanticscholar.org/paper/6bd4845648198b72884d049b2ffc71628f3b837f/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
CzEngVallex - Czech English Valency Lexicon
CzEngVallex - Česko-anglický valenční slovník
CzEngVallex is a bilingual valency lexicon of corresponding Czech and English verbs. It connects 20835 aligned valency frame pairs (verb senses) which are translations of each other, aligning their arguments as well. The CzEngVallex serves as a powerful, real-text-based database of frame-to-frame and subsequently argument-to-argument pairs and can be used for example for machine translation applications. It uses the data from the Prague Czech-English Dependency Treebank project.
CzEngVallex je dvojjazyčný valenční slovník, který obsahuje provázané dvojice českých a anglických sloves. Zahrnuje 20835 odpovídajících slovesných dvojic (překladových ekvivalentů) valenčních rámců (významů slovesa) a zachycuje také propojení jejich argumentů. Tato databáze dvojic rámců a jejich argumentů je založena na reálných textech a může být využita například v aplikacích pro strojový překlad.
[ "Zdeňka Urešová", "Eva Fučíková", "Jan Hajič", "Jana Šindlerová" ]
https://www.semanticscholar.org/paper/08149266b99f7168a51bca479a728396f2fc9c48/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
CzEngVallex: Mapping Valency between Languages
CzEngVallex: Propojování valence mezi jazyky
This report presents a guideline for building a valency lexicon (CzEngVallex) of Czech and English verbs based on the parallel dependency corpus (PCEDT) and two existing valency lexicons: PDT-Vallex a EngVallex.
Technická zprava shrnuje pravidla pro tvorbu valenčního slovníku (CzEngVallexu) českých a anglických sloves, budovaného na základě paralelního závislostního korpusu (PCEDT) a dvou existujících valenčních slovníků: PDT-Vallexu a EngVallexu.
[ "Zdeňka Urešová", "Eva Fučíková", "Jana Šindlerová" ]
https://www.semanticscholar.org/paper/905465545046db7eee2e104af88099bc4b41e52e/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Bilingual English-Czech Valency Lexicon Linked to a Parallel Corpus
Dvojjazyčný anglicko-český valenční slovník propojený s paralelním korpusem
This paper presents a resource and the associated annotation process used in a project of interlinking Czech and English verbal translational equivalents based on a parallel, richly annotated dependency treebank containing also valency and semantic roles, namely the Prague Czech-English Dependency Treebank. One of the main aims of this project is to create a high-quality and relatively large empirical base which could be used both for linguistic comparative research as well as for natural language processing applications, such as machine translation or cross-language sense disambiguation. This paper describes the resulting lexicon, CzEngVallex, and the process of building it, as well some interesting observations and statistics already obtained.
Příspěvek popisuje budování dvojjazyčného valenčního anglicko-českého slovníku CzEngVallex, budovaného na základě paralelního překladového Pražského česko-anglického závislostního korpusu. Díky propojení s korpusovými daty bude možné slovník využívat jak pro lingvistický výzkum, tak pro NLP aplikace.
[ "Zdeňka Urešová", "Ondřej Dušek", "Eva Fučíková", "Jan Hajič", "Jana Šindlerová" ]
https://www.semanticscholar.org/paper/ecbb0475265893e0202b8b805ed392de9b389fa9/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Using the Sketch Engine in Preparation for a Valency Lexicon of Czech Nouns: The Case of Nominalizations of Support Verb Constructions
Využití nástroje Sketch Engine při přípravě valenčního slovníku českých substantiv: Případ nominalizací verbonominálních predikátů s kategoriálním slovesem
I use the Sketch Engine in order to match up collocations standing for particular nominalizations of SVCs. I also search for valency properties of nouns in question using CQL query type which, to some extent, allows searching for syntactic relations in linear corpora. Nominalizations of support verbs display different valency behaviour when compared to notional nouns, avoiding deverbal compounds in English, cf. selling the books / book(-/ )selling / sale of books / books sale vs. selling the idea / *sale of the idea / *idea sale, or loss of memory / memory loss vs. loss of confidence / *confidence loss. In Czech, semantic impoverishing of nominalizations of support verbs has an impact on ways of expressing their Agent, allowing double post-nominal genitives (e.g. sbírání zkušeností nejmladších závodníků ‘gaining experience-GEN.PL [by] youngest-GEN.PL racer-GEN.PL’) which are ungrammatical when the nouns have full semantic meaning. Some nominalized SVCs also allow so-called crossing of nominal valency relations which is distinctive esp. in Czech but also possible in English, cf. giving (opportunity) to the inhabitants vs. opportunity to provide in giving opportunity to the inhabitants to provide for themselves.
Sketch Engine byl využit zejména pro vytipování dvojic substantiv tvořících nominalizace verbonominálních predikátů s kategoriálním slovesem. V prezentaci byly rovněž představeny odlišné syntaktické vlastnosti konstrukcí s plnovýznamovými substantivy a konstrukcí s týmiž substantivy, užitými ve vyprázdněném významu jako nominalizace kategoriálních sloves.
[ "Veronika Kolářová" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
From the Jungle to a Park: Harmonizing Annotations across Languages
Z džungle do parku: Harmonizace anotací napříč jazyky
In this talk I will describe my work towards universal representation of morphology and dependency syntax in treebanks of various languages. Not only is such harmonization advantageous for linguists-users of corpora, it is also a prerequisite for cross-language parser adaptation techniques such as delexicalized parsing. I will present Interset, an interlingua-like tool to translate morphosyntactic representations between tagsets; I will also show how the features from Interset are used in a recent framework called Universal Dependencies. Some experiments with delexicalized parsing on harmonized data will be presented. Finally, I will discuss the extent to which various morphological features are important in the context of statistical dependency parsing.
V této přednášce popíšu svou práci na univerzálním zachycení morfologie a závislostní syntaxe v korpusech různých jazyků. Taková harmonizace je nejen výhodná pro lingvisty-uživatele korpusů, ale je také nezbytným předpokladem pro techniky mezijazykové adaptace parserů, např. delexikalizovaného parsingu. Představím Interset, nástroj podobný interlingvě pro překlad morfosyntaktických reprezentací mezi sadami značek; ukážu také, jak se rysy z Intersetu používají v novém formalismu nazvaném Universal Dependencies. Nakonec proberu důležitost jednotlivých morfologických rysů s ohledem na závislostní parsing.
[ "Daniel Zeman" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Slavic Languages in Universal Dependencies
Slovanské jazyky v Universal Dependencies
We address the application of UD to Slavic languages. We devote the most space to peculiarities of pronouns, determiners, numerals and quantifiers. Other language features that are discussed include modal verbs, ellipsis, nominal predicates, and reflexive pronouns. Most of our examples are from Czech but the language features demonstrated are usually portable to other Slavic languages. We include examples from the other languages where appropriate.
Zabýváme se aplikací UD na slovanské jazyky. Nejvíce prostoru věnujeme zájmenům, determinátorům, číslovkám a kvantifikátorům. Kromě toho diskutujeme i další jazykové jevy jako způsobová slovesa, elipsu, jmenné přísudky a zvratná zájmena. Většina našich příkladů pochází z češtiny, ale jazykové jevy, které na příkladech ukazujeme, jsou obvykle přenositelné na další slovanské jazyky. Tam, kde je to vhodné, uvádíme příklady i z jiných jazyků.
[ "Daniel Zeman" ]
https://www.semanticscholar.org/paper/7a7036da6da5dddc9691c67858866d8f40592b37/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
HamleDT 3.0
HamleDT 3.0
HamleDT (HArmonized Multi-LanguagE Dependency Treebank) is a compilation of existing dependency treebanks (or dependency conversions of other treebanks), transformed so that they all conform to the same annotation style. This version uses Universal Dependencies as the common annotation style.
HamleDT (HArmonized Multi-LanguagE Dependency Treebank) je sbírka existujících závislostních korpusů (nebo do závislostí převedených jiných syntaktických korpusů), transformovaných do jednotného anotačního stylu. Tato verze používá Universal Dependencies jako svůj ústřední anotační styl.
[ "Daniel Zeman", "David Mareček", "Jan Mašek", "Martin Popel", "Loganathan Ramasamy", "Rudolf Rosa", "Jan Štěpánek", "Zdeněk Žabokrtský" ]
https://www.semanticscholar.org/paper/4f7a96246a24b663ff220a12e77da0ba6e5d61ad/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
Machine learning and rule for information extraction
Strojové učení a pravidla pro extrakci informací z textů
As researchers, we decided to develop information extraction systems focused on texts from the legislative and environmental domains. We will present them and we will share the experience on their development. We will focus on the alternatives machine learning vs. rules and raw texts vs. enriched texts. Aspects of their management and evaluation will be discussed as well.
Jako výzkumníci jsme přistoupili k vývoji systémů extrakce informací z textů legislativní a enviromentální domény. V prezentaci systémy představíme a podělíme se o zkušenosti z jejich vývoje. Zaměříme se na alternativy strojové učení vs. pravidla a surové texty vs. texty obohacené o informace. Aspekty správy systémů a způsobu hodnocení jejich kvality budou rovněž diskutovány.
[ "Barbora Hladká" ]
https://www.semanticscholar.org/paper/56631ad651634bbe6309976ca4d33f1b05aebb41/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
A Gentle Introduction to Machine Learning for Natural Language Processing: How to start in 16 practical steps
Úvod do strojového učení v počítačovém zpracování přirozeného jazyka: jak začít v 16 krocích
We present a gentle introduction to machine learning and we explain it on its application in the field of natural language processing. Namely, we address the task of word sense disambiguation practically in the R software system. We focus on readers who approach some task and want to use machine learning, and who do not know how to start.
Prezentujeme úvod do strojového učení prostřednictvím úlohy počítačového zpracování přirozeného jazyka, a sice přiřazení správného významu slov ve větách. Zaměřujeme se hlavně na praktické aspekty v systému R. Článek je urče studentům a mladým vědcům, kteří nemají znalosti strojového učení a chtějí začít.
[ "Barbora Hladká", "Martin Holub" ]
https://www.semanticscholar.org/paper/bf82496f98b0f4e5aecf656b3121efb1700ef686/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
MCLAAS - integrated search system for multilingual archive of testimonies of the Holocaust witnesses
MCLAAS - integrovaný systém vyhledávání ve vícejazyčném audioarchívu výpovědí svědků holocaustu
This functional prototype is used for multi-lingual (Czech and English) search for relevant words or short phrases in the archive of Holocaust survivors, managed by USC (University of Southern California) Shoah Foundation Institute (http://dornsife.usc.edu/vhi/), which contains more than 110,000 hours of records in 32 languages, with approximately half of these interviews is conducted in English. Czech part of the archive accounts for approximately one thousand hours. For the purposes of searching in the system MCLAAS are Czech and English speech data first processed with the appropriate speech recognition module (SEASR-CZE - see http://www.kky.zcu.cz/en/sw/SEASR-CZE or SEASR-ENG - see http://www.kky.zcu.cz/en/sw/SEASR-ENG) and then a so-called index is created, which is a machine representation of recognized utterances, which speeds up the search for a desired word or phrase. Those are all data available in the case of Czech; in English there are more than 50,000 hours, but the recognition and indexing of all this volume will require parallelization of individual processes. Cross-searching (query in English, data / interviews in English and Czech) in the system is facilitated by automatic query translation. Implementation was carried out as a special version of MTMonkey (http://ufal.mff.cuni.cz/mtmonkey).
Tento funkční vzorek slouží pro vícejazyčné (české a anglické) vyhledávání relevantních slov či krátkých frází v archivu přeživších Holocaustu, spravovaném USC (University of Southern California) Shoah Foundation Institute (http://dornsife.usc.edu/vhi/). Tento archiv obsahuje více než 110 tisíc hodin záznamů v 32 jazycích, přičemž přibližně polovina těchto rozhovorů je vedena v angličtině. Česká část archivu obnáší zhruba jeden tisíc hodin. Funkční vzorek se skládá ze serverového počítače, softwarových modulů MCLASS (http://www.kky.zcu.cz/cs/sw/MCLAAS), WFBAS (http://www.kky.zcu.cz/cs/sw/WFBAS), pracovní databáze sestavené softwary SEASR-CZE (http://www.kky.zcu.cz/cs/sw/SEASR-CZE) a SEASR-ENG (http://www.kky.zcu.cz/cs/sw/SEASR-ENG) a tenkého klienta s obvyklým webovým prohlížečem. Serverový počítač je počítač s konfigurací odpovídající náročnosti vykonávané úlohy s připojením k internetu. Počítač použitý pro funkční vzorek má 2 procesory Intel(R) Xeon(R) CPU E5-2620 v2 @ 2.10GHz. Pro účely vyhledávání v systému jsou česká a anglická řečová data nejprve zpracována příslušným modulem rozpoznávání řeči (SEASR-CZE, resp. SEASR-ENG).. Každý ze systémů v současnosti hledá výskyty slov či frází zhruba v 1000 hodin videozáznamů. V případě češtiny jde o veškerá dostupná data; v angličtině je k dispozici více než 50 tisíc hodin, ale rozpoznání a zaindexování celého tohoto objemu bude vyžadovat paralelizaci jednotlivých procesů. Pro křížové vyhledávání (dotaz v češtině, data/rozhovory v angličtině a češtině) v softwaru byl použit systém překladu dotazu. Implementace byla provedena jako zvláštní verze systému MTMonkey (http://ufal.mff.cuni.cz/mtmonkey)
[ "Jan Švec", "Adam Chýlek", "Petr Stanislav", "Pavel Ircing", "Aleš Pražák", "Josef V. Psutka", "Petr Salajka", "Jan Zelinka", "Lucie Skorkovská", "Zbyněk Zajíc", "Daniel Soutner", "Luděk Müller", "Ondřej Bojar", "Aleš Tamchyna", "David Mareček", "Zdeněk Žabokrtský", "Martin Popel", "Petra Galuščáková", "Zdeňka Urešová", "Jan Hajič" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Free or Fixed Word Order: What Can Treebanks Reveal?
Volný nebo pevný slovosled: Co zjistíme z treebanku?
The paper describes an ongoing experiment consisting in the attempt to quantify word-order properties of three Indo-European languages (Czech, English and German). The statistics are collected from the syntactically annotated treebanks available for all three languages. The treebanks are searched by means of a universal query tool PML-TQ. The search concentrates on the mutual order of a verb and its complements (subject, object(s)) and the statistics are calculated for all permutations of the three elements. The results for all three languages are compared and a measure expressing the degree of word order freedom is suggested in the final section of the paper. This study constitutes a motivation for formal modeling of natural language processing methods.
Článek popisuje probíhající experiment soustřeďující se na kvantifikaci slovosledných vlastností tří indoevropských jazyků, češtiny, angličtiny a němčiny. Statistiky jsou získávány ze syntakticky anotovaných korpusů za pomoci dotazovacího jazyka PML-TQ. tato studie slouží jako motivace pro formální modelování metod NLP.
[ "Vladislav Kuboň", "Markéta Lopatková" ]
https://www.semanticscholar.org/paper/42d3fd56534a1fcc16939fdd134fbd679989da4c/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Word-Order Analysis Based Upon Treebank Data
Slovosledná analýza na základě korpusových dat
The paper describes an experiment consisting in the attempt to quantify word-order properties of three Indo-European languages (Czech, English and Farsi). The investigation is driven by the endeavor to find an objective way how to compare natural languages from the point of view of the degree of their word-order freedom. Unlike similar studies which concentrate either on purely linguistic or purely statistical approach, our experiment tries to combine both - the observations are verified against large samples of sentences from available treebanks, and, at the same time, we exploit the ability of our tools to analyze selected important phenomena (as, e.g., the differences of the word order of a main and a subordinate clause) more deeply. The quantitative results of our research are collected from the syntactically annotated treebanks available for all three languages. Thanks to the HamleDT project, it is possible to search all treebanks in a uniform way by means of a universal query tool PML-TQ. This is also a secondary goal of this paper - to demonstrate the research potential provided by language resources which are to a certain extent unified.
Článek popisuje experiment soustřeďující se na zkoumání a kvantifikaci slovosledných vlastností na základě dat dostupných díky projektu HamleDT. Tato studie usiluje o objektivní porovnání přirozených jazyků, a to kombinováním jak hledisek čistě lingvistických, tak hledisek kvantitativních, která jsou založena na velkém souboru dostupných dat. Tato studie slouží jako motivace pro formální modelování metod NLP; zároveň si klade za cíl představit výzkumný potenciál jazykových zdrojů se standardizovanou anotací.
[ "Vladislav Kuboň", "Markéta Lopatková" ]
https://www.semanticscholar.org/paper/2e702aa576630895c79b5a6b3b61fd2e5624cda0/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Analysis of Coordinating Constructions in a Dependency Treebank
Analýza koordinačních konstrukcí v závislostním korpusu
This paper summarizes results of automatic analysis of coordinating constructions and appositions in the Prague Dependency Treebank using a method of analysis by reduction. Experiments are performed on a large subset of the treebank. This subset is obtained as a result of a query providing a set of more than 4,300 suitable sentences and their tree structures containing coordinations and appositions. The automatic procedure is complemented by a manual analysis of reasons why certain sentences (trees) were not fully reduced. This analysis helps to gain a better insight into the phenomena of co- ordination and apposition and their formal properties.
Tento článek shrnuje experimenty týkající se automatické analýzy koordinačních a apozičních konstrukcí v Pražském závislostním korpuse na základě metody redukční analýzy.
[ "Vladislav Kuboň", "Markéta Lopatková", "Jiří Mírovský" ]
https://www.semanticscholar.org/paper/9937d608dacda1c5a7e4830da7aa1955ed4a22a1/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
Automatic linking of lexicographic sources and corpus data
Automatické propojování lexikografických zdrojů a korpusových dat
Along with the increasing development of language resources – i.e., new lexicons, lexical databases, corpora, treebanks – the need for their efficient interlinking is growing. With such a linking, one can easily benefit from all their properties and information. Considering the convergence of resources, universal lexicographic formats are frequently discussed. In the present thesis, we investigate and analyse methods of interlinking language resources automatically. We introduce a system for interlinking lexicons (such as VALLEX, PDT-Vallex, FrameNet or SemLex) that offer information on syntactic properties of their entries. The system is automated and can be used repeatedly with newer versions of lexicons under development. We also design a method for identification of multiword expressions in a parsed text based on syntactic information from the SemLex lexicon. An output that verifies feasibility of the used methods is, among others, the mapping between the VALLEX and the PDT-Vallex lexicons, resulting in tens of thousands of annotated treebank sentences from the PDT and the PCEDT treebanks added into VALLEX.
Spolu se vznikem stále dalších jazykových zdrojů – slovníků, lexikálních databází, korpusů, treebanků – roste i potřeba jejich účinného propojování, které by umožnilo snadné využití veškerých shromážděných vlastností a informací. V tomto ohledu je také aktuální téma univerzálních lexikografických formátů. Tato práce zkoumá metody automatického propojování jazykových dat. Představíme zde systém na propojování slovníků, jakými jsou například VALLEX, PDT-Vallex, FrameNet, nebo SemLex, které poskytují syntaktickou informaci o svých heslech. Systém je automatický, umožňuje tudíž opakovanou aplikaci na novější verze vyvíjejících se jazykových zdrojů. Na základě syntaktické informace obsažené ve slovníku víceslovných výrazů SemLex navrhujeme metodu vyhledávající tyto výrazy v automaticky anotovaném textu. Praktickým výstupem potvrzujícím úspěšnost použitých metod je mj. propojení slovníků VALLEX a PDT-Vallex vedoucí k doplnění desítek tisíc anotovaných vět z treebanků PDT a PCEDT do VALLEXu.
[ "Eduard Bejček" ]
https://www.semanticscholar.org/paper/28a93338086418b14591fa58a6b79e116c99efc9/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
Linking valency lexicons
Propojování valenčních slovníků
Linking Czech valency lexicons VALLEX and PDT-Vallex will connect VALLEX with tens of thousands of example sentences in corpus data of PDT. The presentation will introduce the data format and the approach used to automatick linking and the published results.
Spojení slovníků na úrovni hesel či jejich částí přináší zpravidla rozšíření informací pro obě strany. Spojení českých valenčních slovníků VALLEX a PDT-Vallex přinese VALLEXu napojení na data PDT a tudíž desítky tisíc příkladových vět; PDT-Vallexu pak další syntakticko-sémantické informace (jako je reciprocita, kontrola, sémantická třída, ad.). Referát představí datový formát a postup použitý při propojování lexikálních jednotek a dosažené a zveřejněné výsledky.
[ "Eduard Bejček" ]
https://www.semanticscholar.org/paper/e57f7423fdcb39c1b3c4d094306051d19a8c7bfc/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Identification of Multiword BBN Named Entities in Dependency Annotation of Wall Street Journal
Identifikace víceslovných pojmenovaných entit BBN v závislostní anotaci Wall Street Journalu
We compiled a lexicon of multiword BBN entities and used it for consistency checking of annotations (WSJ, BBN and PDT annotations).
Sestavili jsme slovník víceslovných entit BBN a následně jsme jej využili ke kontrolám anotací (anotace WSJ, BBN, PCEDT).
[ "Eduard Bejček", "Pavel Straňák" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Morphology within the Multi-Layered Annotation Scenario of the Prague Dependency Treebank
Morfologie ve vícerovinném anotačním scénáři Pražského závislostního korpusu
After a brief introduction into the rich morphology of the Czech language, I focus on the multi-layered annotation scheme of the Prague Dependency Treebank. Morphological categories are described at a separate annotation layer. The meanings conveyed by this categories are involved in the deep-syntatic annotation layer as well. The semiautomatic annotation is illustrated by several examples in the talk. Development of Czech taggers, named entity recognition tools and our recent work on a resource of derivational morphology are outlined as a few topics based on Praguian morphology.
Po krátkém úvodu ilustrujícím bohatost české morfologie se v přednášce zaměřuji na způsob zachycení morfologických kategorií v Pražském závislostním korpusu. V druhé části přednášky se věnuji přehledu historie taggerů češtiny, rozpoznávání pojmenovaných entit v češtině a naší aktuální práci na budování lexikálního zdroje derivační morfologie češtiny.
[ "Magda Ševčíková" ]
https://www.semanticscholar.org/paper/725d9538ff6b7b50477db721441f74022597f5e2/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Morphology within the Multi-Layered Annotation Scenario of the Prague Dependency Treebank
Morfologie ve vícerovinném anotačním scénáři Pražského závislostního korpusu
Morphological annotation constitutes a separate layer in the multilayered annotation scenario of the Prague Dependency Treebank. At this layer, morphological categories expressed by a word form are captured in a positional part-of-speech tag. According to the Praguian approach based on the relation between form and function, functions (meanings) of morphological categories are represented as well, namely as grammateme attributes at the deep-syntactic (tectogrammatical) layer of the treebank. In the present paper, we first describe the role of morphology in the Prague Dependency Treebank, and then outline several recent topics based on Praguian morphology: named entity recognition in Czech, formemes attributes encoding morpho-syntactic information in the dependency-based machine translation system, and development of a lexical database of derivational relations based partially on information provided by the morphological analyser.
Slovnědruhová kategorie a morfologické rysy tokenů jsou v Pražském závislostním korpusu zachyceny na zvláštní anotační rovině, významy vyjadřované morfologickými kategorie jsou ovšem také součástí hloubkově syntaktické roviny. V příspěvku podrobně představujeme zachycení morfologických kategorií v tomto korpusu a věnujeme se také aplikacím, pro které byla tato morfologicky anotovaná data použita.
[ "Magda Ševčíková" ]
https://www.semanticscholar.org/paper/725d9538ff6b7b50477db721441f74022597f5e2/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
Lexikal Network DeriNet
Lexikální síť DeriNet
The presentation will be focused on the lexical NetWork DeriNet that captures morphological derivations in Czech.
Na přednášce bude představena lexikální síť DeriNet, která zachycuje derivační vztahy v morfologii češtiny.
[ "Magda Ševčíková", "Zdeněk Žabokrtský" ]
https://www.semanticscholar.org/paper/d0458a8f52f2bfc17813bce341e0bad11b3960b0/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Universal Dependencies 1.0
Universal Dependencies 1.0
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).
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).
[ "Joakim Nivre", "Cristina Bosco", "Jinho Choi", "Marie-Catherine de Marneffe", "Timothy Dozat", "Richárd Farkas", "Jennifer Foster", "Filip Ginter", "Yoav Goldberg", "Jan Hajič", "Jenna Kanerva", "Veronika Laippala", "Alessandro Lenci", "Teresa Lynn", "Christopher Manning", "Ryan McDonald", "Anna Missilä", "Simonetta Montemagni", "Slav Petrov", "Sampo Pyysalo", "Natalia Silveira", "Maria Simi", "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
Universal Dependencies 1.2
Universal Dependencies 1.2
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 third release of UD Treebanks, Version 1.2.
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 třetí vydání treebanků UD, verze 1.2.
[ "Joakim Nivre", "Željko Agić", "Maria Jesus Aranzabe", "Masayuki Asahara", "Aitziber Atutxa", "Miguel Ballesteros", "John Bauer", "Kepa Bengoetxea", "Riyaz Ahmad Bhat", "Cristina Bosco", "Sam Bowman", "Giuseppe G. A. Celano", "Miriam Connor", "Marie-Catherine de Marneffe", "Arantza Diaz de Ilarraza", "Kaja Dobrovoljc", "Timothy Dozat", "Tomaž Erjavec", "Richárd Farkas", "Jennifer Foster", "Daniel Galbraith", "Filip Ginter", "Iakes Goenaga", "Koldo Gojenola", "Yoav Goldberg", "Berta González Saavedra", "Bruno Guillaume", "Jan Hajič", "Dag Haug", "Radu Ion", "Elena Irimia", "Anders Johannsen", "Hiroshi Kanayama", "Jenna Kanerva", "Simon Krek", "Veronika Laippala", "Alessandro Lenci", "Nikola Ljubešić", "Teresa Lynn", "Christopher Manning", "Cătălina Mărănduc", "David Mareček", "Héctor Martínez Alonso", "Jan Mašek", "Yuji Matsumoto", "Ryan McDonald", "Anna Missilä", "Verginica Barbu Mititelu", "Yusuke Miyao", "Simonetta Montemagni", "Shinsuke Mori", "Hanna Nurmi", "Petya Osenova", "Lilja Øvrelid", "Elena Pascual", "Marco Passarotti", "Cenel-Augusto Perez", "Slav Petrov", "Jussi Piitulainen", "Barbara Plank", "Martin Popel", "Prokopis Prokopidis", "Sampo Pyysalo", "Loganathan Ramasamy", "Rudolf Rosa", "Shadi Saleh", "Sebastian Schuster", "Wolfgang Seeker", "Mojgan Seraji", "Natalia Silveira", "Maria Simi", "Radu Simionescu", "Katalin Simkó", "Kiril Simov", "Aaron Smith", "Jan Štěpánek", "Alane Suhr", "Zsolt Szántó", "Takaaki Tanaka", "Reut Tsarfaty", "Sumire Uematsu", "Larraitz Uria", "Viktor Varga", "Veronika Vincze", "Zdeněk Žabokrtský", "Daniel Zeman", "Hanzhi Zhu" ]
https://www.semanticscholar.org/paper/1e105a47662affe083eaf110e5590ad4b9f7894d/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
A new parsing algorithm
Nový algoritmus pro syntaktickou analýzu
We propose a new dependency parsing algorithm, designed to allow for any features while maintaining tractability. The main idea is to start with a baseline parse tree rather than with no tree at all, and to transform that tree into the correct one by repeated application of simple transformation operations. We focus on inference, discussing anticipated issues and possible remedies. Suggestions towards the training procedure and the feature set are also briefly presented.
V článku je navržen nový algoritmus pro závislostní syntaktickou analýzu, který umožňuje použít různé typy predikčních rysů a přitom zůstává výpočetně zvládnutelný. Základní myšlenka vychází z postupného upravování závislostního stromu opakovaným aplikovaním jednoduchých transformačních operací.
[ "Rudolf Rosa" ]
https://www.semanticscholar.org/paper/fdf2c8bd6afe0b4ef9265d3575e62b75c173b018/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Depfix: Automatic Post-editing of SMT
Depfix: Automatická post-editace SSP
Depfix, an open-source system for automatic post-editing of phrase-based machine translation outputs. Depfix employs a range of natural language processing tools to obtain analyses of the input sentences, and uses a set of rules to correct common or serious errors in machine translation outputs.
Depfix -- open-souce systém pro automatickou post-editaci výstupů frázového strojového překladu. Depfix zapojuje řadu nástrojů pro automatické zpracování přirozeného jazyka, pomocí nichž získává rozbor vstupních vět, a používá sadu pravidel pro opravu závažných chyb či chyb obvyklých ve výstupech strojových překladačů.
[ "Rudolf Rosa" ]
https://www.semanticscholar.org/paper/756c0eb3b2c4b9e6023d999ce31ae2d4f94e44f0/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
MSTperl delexicalized parser transfer scripts and configuration files
Skripty pro přenos delexikalizovaného parseru MSTperl a soubory konfigurace
This is a set of MSTperl parser configuration files and scripts for delexicalized parser transfer. They were used in the work reported in arXiv:1506.04897 (http://arxiv.org/abs/1506.04897), as well as several related papers. The MSTperl parser is available at http://hdl.handle.net/11234/1-1480
Toto je sada konfiguračních souborů parseru MSTperl a skriptů pro přenos delexikalizovaného parseru. Byly použity v práci popsané v článku arXiv:1506.04897 (http://arxiv.org/abs/1506.04897) a v několika souvisejících článcích. Parser MSTperl je dostupný na adrese http://hdl.handle.net/11234/1-1480
[ "Rudolf Rosa" ]
https://www.semanticscholar.org/paper/469e53126d3963b0ba9f5b94c2c2e8ddab100e5f/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
MSTperl parser (2015-05-19)
MSTperl parser (2015-05-19)
MSTperl is a Perl reimplementation of the MST parser of Ryan McDonald, with several additional advanced functions, such as support for parallel features.
MSTperl je Perlovou reimplementací MST parseru Ryana McDonalda, s několika pokročilými funkcemi navíc, jako je podpora pro paralelní rysy.
[ "Rudolf Rosa" ]
https://www.semanticscholar.org/paper/3bee60095259fcbadea1e03fc74aba90177d52f3/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Multi-source Cross-lingual Delexicalized Parser Transfer: Prague or Stanford?
Vícezdrojový mezijazyčný přenos delexikalizovaného parseru: Praha nebo Stanford?
We compare two annotation styles, Prague dependencies and Universal Stanford Dependencies, in their adequacy for parsing. We specifically focus on comparing the adposition attachment style, used in these two formalisms, applied in multi-source cross-lingual delexicalized dependency parser transfer performed by parse tree combination. We show that in our setting, converting the adposition annotation to Stanford style in the Prague style training treebanks leads to promising results. We find that best results can be obtained by parsing the target sentences with parsers trained on treebanks using both of the adposition annotation styles in parallel, and combining all the resulting parse trees together after having converted them to the Stanford adposition style (+0.39% UAS over Prague style baseline). The score improvements are considerably more significant when using a smaller set of diverse source treebanks (up to +2.24% UAS over the baseline).
Porovnáváme dva anotační styly, Pražské závislosti a univerzální Stanfordské závislosti, ve smyslu jejich vhodnosti pro parsing. Konkrétně se zaměřujeme na porovnání stylu zavěšení adpozic, použivaného v těchto dvou formalismech, na úloze vícezdrojového mezijazyčného přenosu delexikalizovaného parseru, používajíce MSTParser. Zjišťujeme, že v našem scénáři se stává zřetelnou výhoda Stanfordského stylu, neboť převod anotace adpozic v treebancích anotovaných v Pražském stylu do Stanfordského stylu vede k mírně lepšímu výsledku (+0.2% UAS). Dále ukazujeme, že nejlepších výsledků lze dosáhnout pomocí natrénování parserů na treebancích využívajících oba styly anotace adpozic, analýzy cílového treebanku pomocí všech těchto parserů a kombinace všech získaných stromů, po jejich převodu do stejného anotačního stylu (dalších +0.18% UAS). Rozdíly ve skóre jsou ještě vyšší, když se použije menší sada různorodých zdrojových treebanků (až 2.24% UAS oproti základní verzi).
[ "Rudolf Rosa" ]
https://www.semanticscholar.org/paper/02aa30f2e68a95971b9ca1d525a97aeb881df049/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Parsing Natural Language Sentences by Semi-supervised Methods
Syntaktická analýza vět přirozeného jazyka pomocí částečně řízených metod
We present our work on semi-supervised parsing of natural language sentences, focusing on multi-source crosslingual transfer of delexicalized dependency parsers. We first evaluate the influence of treebank annotation styles on parsing performance, focusing on adposition attachment style. Then, we present KLcpos3, an empirical language similarity measure, designed and tuned for source parser weighting in multi-source delexicalized parser transfer. And finally, we introduce a novel resource combination method, based on interpolation of trained parser models.
Představujeme naši práci v oblasti částečně řízené syntaktické analýzy vět přirozeného jazyka, se zaměřením na mezijazyčný přenos delexicalizovaných závislostních parserů s více zdroji. Nejprve vyhodnocujeme vliv anotačního stylu treebanku na úspěšnost parsingu, se zaměřením na styl zavěšení adpozic. Poté představujeme KLcpos3, empirickou míru podobnosti jazyků, navrženou a vyladěnou pro vážení zdrojových parserů při přenosu delexicalizovaných parserů s více zdroji. Nakonec představíme novou metodu kombinace zdrojů, založenou na interpolaci natrénovaných modelů parserů.
[ "Rudolf Rosa" ]
https://www.semanticscholar.org/paper/62463de99a24602dc145bb017db23cfa3b0690ff/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Translation Model Interpolation for Domain Adaptation in TectoMT
Interpolace překladových modelů pro doménovou adaptaci v TectoMT
We present an implementation of domain adaptation by translation model interpolation in the TectoMT translation system with deep transfer. We evaluate the method on six language pairs with a 1000-sentence in-domain parallel corpus, and obtain improvements of up to 3 BLEU points. The interpolation weights are set uniformly, without employing any tuning.
Představujeme implementaci doménové adaptace pomocí interpolace překladových modelů v TectoMT, překladovém systému s hloubkovým transferem. Vyhodnotili jsme tuto metodu na šesti jazykových párech s doménovým paralelním korpusem o 1000 větách, a získali jsme zlepšení až o 3 body BLEU. Váhy pro interpolaci jsou nastaveny uniformně, bez zapojení jakéhokoliv ladění.
[ "Rudolf Rosa", "Ondřej Dušek", "Michal Novák", "Martin Popel" ]
https://www.semanticscholar.org/paper/94f334b90057bc985963e9a9efc2a493c0179dcd/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
KLcpos3 - a Language Similarity Measure for Delexicalized Parser Transfer
KLcpos3 - míra podobnosti jazyků pro delexicalizovaný přenos parseru
We present KLcpos3, a language similarity measure based on Kullback-Leibler divergence of coarse part-of-speech tag trigram distributions in tagged corpora. It has been designed for multilingual delexicalized parsing, both for source treebank selection in single-source parser transfer, and for source treebank weighting in multi-source transfer. In the selection task, KLcpos3 identifies the best source treebank in 8 out of 18 cases. In the weighting task, it brings +4.5% UAS absolute, compared to unweighted parse tree combination.
Představujeme KLcpos3, míru podobnosti jazyků založenou na Kullbackově-Leiblerově divergenci rozložení trigramů hrubých značek slovních druhů v otagovaných korpusech. Tato míra byla navržena pro vícejazyčný delexicalizovaný parsing, a to jak pro výběr zdrojového treebanku při přenosu parseru s jedním zdrojem, tak pro vážení zdrojových treebanků při přenosu parseru s více zdroji. V úloze výběru zdroje rozpozná KLcpos3 nejlepší zdrojový treebank v 8 z 18 případů. V úloze vážení zdroje přínáší zvýšení UAS o +4.5 procentního bodu oproti nevážené kombinaci stromů.
[ "Rudolf Rosa", "Zdeněk Žabokrtský" ]
https://www.semanticscholar.org/paper/f54adda402157425a544f25ac7ff7cdae053ba9a/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
MSTParser Model Interpolation for Multi-source Delexicalized Transfer
Interpolace modelů MSTParseru pro vícezdrojový delexikalizovaný přenos
We introduce interpolation of trained MSTParser models as a resource combination method for multi-source delexicalized parser transfer. We present both an unweighted method, as well as a variant in which each source model is weighted by the similarity of the source language to the target language. Evaluation on the HamleDT treebank collection shows that theweightedmodelinterpolationperforms comparably to weighted parse tree combination method, while being computationally much less demanding.
Představujeme interpolaci natrénovaných modelů MSTParseru jako metodu kombinace zdrojů pro vícezdrojový delexikalizovaný přenos parseru. Představujeme jak neváženou metodu, tak variantu ve které je každý zdrojový model vážen podobností zdrojového a cílového jazyka. Vyhodnocení na sbírce treebanků HamleDT ukazuje, že vážená interpolace modelů má podobnou úspěšnost jako vážená kombinace stromů, přičemž je mnohem méně komputačně náročná.
[ "Rudolf Rosa", "Zdeněk Žabokrtský" ]
https://www.semanticscholar.org/paper/5e4efddd43f6c3e3fc36cb91f083352d93f7eed1/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Delving deep into some UFAL data and tools
Seznamování se s daty a softwarovými nástroji ÚFALu
This paper is devoted to three topics concerning my first year postgraduate studies at UFAL. First, I will present the current state-of-the-art of coreference resolution in the Czech and English language. Next, I will describe a method of extraction of numerical expressions linked to certain entities with the GATE tool and the last part will be dedicated to the Capek sentence diagramming editor and a method of getting more annotated data from sentence diagrams.
Článek je zaměřen na tři témate související s autorčiným prvním rokem doktorského studia. Zaprvé, zabývá se nástroji pro rozpoznávání koreference v češtině a angličtině. Zadruhé, popisuje metody extrakce číselných výrazů napojených na vybrané typy entit v prostředí GATE. Poslední část je věnovaná editoru Capek a získávání anotovaných větných diagramů
[ "Ivana Lukšová" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Comparison of Coreference Resolvers for Deep Syntax Translation
Porovnání systémů na rozpoznávaní koreference v překladu přes hloubkovou syntax
This work focuses on using anaphora for machine translation with deep-syntactic transfer. We compare multiple coreference resolvers for English in terms of how they affect the quality of pronoun translation in English-Czech and English-Dutch machine translation systems with deep transfer. We examine which pronouns in the target language depend on anaphoric information, and design rules that take advantage of this information. The resolvers’ performance measured by translation quality is contrasted with their intrinsic evaluation results. In addition, a more detailed manual analysis of English-to-Czech translation was carried out.
Tato práce porovnává několik systémů pro rozpoznávání koreference v angličtině v překladu přes hloubkovou syntax z angličtiny do češtiny a holandštiny.
[ "Michal Novák", "Dieke Oele", "Gertjan van Noord" ]
https://www.semanticscholar.org/paper/250344091c719cb4a003e347bbff564de9cd9502/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Coreferential expressions in English and Czech
Koreferenční výrazy v češtině a angličtině
In this talk, we present a comprehensive study on mappings between certain classes of coreferential expressions in English and Czech. We focused on central pronouns, relative pronouns and anaphoric zeros. For instance, the English sentence "It switched to a caffeine-free formula using its new Coke in 1985" has been in PCEDT translated to "V roce 1985 přešla na bezkofeinovou recepturu, kterou používá pro svojí novou kolu". This pair of sentences exhibits several types of changes in expressing coreference: English personal pronouns turns into a Czech zero, possessive pronoun into a possessive reflexive and finally, the -ing participle has been translated to a relative clause. In a similar manner, we have collected a statistics of mappings from a subsection of PCEDT, which we will support by multiple examples and contrast with the theoretical assumptions. For such a study, the quality of word alignment is crucial. Thus, we designed a rule-based refining algorithm for English personal and possessive pronouns and Czech relative pronouns, which served as an automatic alignment pre-annotation. Subsequently, this annotation has been manually corrected and completed, obtaining a basis for this empirical study.
V teto prednasce predstavujeme srovnani koreferencnich vyrazu v cestine a anglictine na zaklade paralelnich dat Prazskeho anglicko-ceskeho zavislostniho korpusu. V centru naseho vyzkumu jsou osobni, privlastnovaci, reflexivni a vztazna zajmena. Zvlastni pozornost venujeme rovnez vynechani zajmen v anaforicke pozici.
[ "Michal Novák", "Anna Nedoluzhko" ]
https://www.semanticscholar.org/paper/00409c5fff8e4bf5ece7ae9c4821e3b6dbc9e1da/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Correspondences between Czech and English Coreferential Expressions
Vztahy mezi českými a anlgličkými koreferenčními výrazy
V této práci představujeme podrobný srovnávací výzkum některých typů koreferenčních vztahů v češtině a angličtině na paralelním česko-anglickém korpusu. Zaměřujeme se na centrální, vztažná a anaforická zájmena. Vypracovali jsme algoritmus vylepšení automatického vyrovnání několika typů českých a anglických zájmen, který vylepšuje kvalitu vyrovnání nejen pro tyto typy ale a pro všechny ostatní. Korespondence daných typů zájmen byly anatovány ručně a dále lingvisticky analyzovány.
In this work, we present a comprehensive study on correspondences between certain classes of coreferential expressions in English and Czech. We focus on central pronouns, relative pronouns, and anaphoric zeros. We designed an alignment-refining algorithm for English personal and possessive pronouns and Czech relative pronouns that improves the quality of alignment links not only for the classes it aimed at but also in general. Moreover, the instances of anaphoric expressions we focus on were manually annotated with their alignment counterparts, which served as a basis for this empirical study. The collected statistics of correspondences are contrasted with theoretical assumptions regarding the use of anaphoric means in the languages under analysis, such as pro-drop properties, the use of finite and non-finite constructions, etc. Finally, we present the ways how the observed correspondences can be exploited in cross-lingual coreference resolution.
[ "Michal Novák", "Anna Nedoluzhko" ]
https://www.semanticscholar.org/paper/00409c5fff8e4bf5ece7ae9c4821e3b6dbc9e1da/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
Comparison of coreferential expressions in Czech and English
Srovnání koreferenčních výrazů v češtině a angličtině
In this work, we present a comprehensive study on correspondences between certain classes of coreferential expressions in English and Czech. We focus on central pronouns, relative pronouns, and anaphoric zeros. We designed an alignment-refining algorithm for English personal and possessive pronouns and Czech relative pronouns that improves the quality of alignment links not only for the classes it aimed at but also in general. Moreover, the instances of anaphoric expressions we focus on were manually annotated with their alignment counterparts, which served as a basis for this empirical study. The collected statistics of correspondences are contrasted with theoretical assumptions regarding the use of anaphoric means in the languages under analysis, such as pro-drop properties, the use of finite and non-finite constructions, etc. Finally, we present the ways how the observed correspondences can be exploited in cross-lingual coreference resolution.
V této práci představujeme podrobný srovnávací výzkum některých typů koreferenčních vztahů v češtině a angličtině na paralelním česko-anglickém korpusu. Zaměřujeme se na centrální, vztažná a anaforická zájmena. Vypracovali jsme algoritmus vylepšení automatického vyrovnání několika typů českých a anglických zájmen, který vylepšuje kvalitu vyrovnání nejen pro tyto typy ale a pro všechny ostatní. Korespondence daných typů zájmen byly anatovány ručně a dále lingvisticky analyzovány.
[ "Michal Novák", "Anna Nedoluzhko" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
MT-ComparEval: Graphical evaluation interface for Machine Translation development
MT-ComparEval: grafický nástroj pro vyhodnocování vývoje strojového překladu
The tool described in this article has been designed to help MT developers by implementing a web-based graphical user interface that allows to systematically compare and evaluate various MT engines/experiments using comparative analysis via automatic measures and statistics. The evaluation panel provides graphs, tests for statistical significance and n-gram statistics. We also present a demo server http://wmt.ufal.cz with WMT14 and WMT15 translations.
Nástroj MT-ComparEval byl navržen tak, aby umožnil vývojářům strojového překladu porovnávat a vyhodnocovat různé MT systémy a jejich verze. MT-ComparEval obsahuje několik metrik pro vyhodnocení strojového překladu.
[ "Ondřej Klejch", "Eleftherios Avramidis", "Aljoscha Burchardt", "Martin Popel" ]
https://www.semanticscholar.org/paper/9ba19eab999cefef4dc29f1e6a9046458b09a7fb/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
CloudASR - v1.0
CloudASR - v1.0
CloudASR is a cloud platform for automatic speech recognition, which supports batch and online speech recognition mode. Its main features are scalability, customizability and easy deployment.
CloudASR je cloudová platforma pro automatické rozpoznávání řeči, která umožňuje dávkové i online zpracování nahrávek. Její hlavní přednosti jsou škálovatelnost, přizpůsobitelnost a snadný proces nasazaní.
[ "Ondřej Klejch", "Ondřej Plátek", "Lukáš Žilka", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/eb82e449285f8b23cf86fc20dcac89bb265d0339/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
CloudASR: Platform & Service
CloudASR: Platforma & Služba
CloudASR is a software platform and a public ASR web-service. Its three strong features are state-of-the-art online speech recognition performance, easy deployment, and scalability. Furthermore, it contains an annotation interface for the addition of transcriptions for the recordings. The platform API supports both batch and online speech recognition. The batch version is compatible with Google Speech API. New ASR engines can be added onto the platform and can work simultaneously.
CloudASR je softwarová platforma a veřejná služba pro rozpoznávání řeči. Její tři silné stránky jsou: kvalita rozpoznávání na úrovni stavu poznání, jednoduché nasazení, a škálovatelnost. Dále, obsahuje anotační rozhraní pro přidávání transkripcí. API platformy podporuje jak dávkovou tak online rozpoznávání. Dávková verze je kompatibilní s Google Speech API. Platforma umožňuje přidání nových rozpoznávačů, které potom můžou fungovat paralelně vedle sebe.
[ "Ondřej Klejch", "Ondřej Plátek", "Lukáš Žilka", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/70783aef32ac4e4e99f3a8ebf4d9835265957918/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
MT-ComparEval
MT-ComparEval
MT-ComparEval is a tool for Machine Translation developers, which allows to compare and evaluate different MT systems (and their versions). MT-ComparEval includes several automatic MT evaluation metrics.
MT-ComparEval je nástroj, který umožňuje vývojářům strojového překladu porovnávat a vyhodnocovat různé MT systémy a jejich verze. MT-ComparEval obsahuje několik metrik pro vyhodnocení strojového překladu.
[ "Ondřej Klejch", "Martin Popel", "Roman Sudarikov" ]
https://www.semanticscholar.org/paper/9ba19eab999cefef4dc29f1e6a9046458b09a7fb/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Improvements to Korektor: A case study with native and non-native Czech
Zlepšení korektor: Případová studie s nativní a non-rodilého Čecha
We present recent developments of Korektor, a statistical spell checking system. In addition to lexicon, Korektor uses language models to find real-word errors, detectable only in context. The models and error probabilities, learned from error corpora, are also used to suggest the most likely corrections. Korektor was originally trained on a small error corpus and used language models extracted from an in-house corpus WebColl. We show two recent improvements: • We built new language models from freely avail- able (shuffled) versions of the Czech National Corpus and show that these perform consistently better on texts produced both by native speakers and non-native learners of Czech. • We trained new error models on a manually annotated learner corpus and show that they perform better than the standard error model (in error detection) not only for the learners’ texts, but also for our standard eval- uation data of native Czech. For error correction, the standard error model outperformed non-native models in 2 out of 3 test datasets. We discuss reasons for this not-quite-intuitive improvement. Based on these findings and on an analysis of errors in both native and learners’ Czech, we propose directions for further improvements of Korektor.
Představujeme aktuální vývoj korektor, je statistický systém kontroly pravopisu. Kromě lexikonu, Korektor používá jazyk modely najít chyby real-slovo, detekovatelná pouze v kontextu. Modely a chyba probanického, vyvozené z chyb korpusů, jsou také používány pro navrhovaly GEST nejpravděpodobnější opravy. Korektor byl původně vyškolení na malé chyby korpusu a použité jazykové modely extrahuje z in-house corpus WebColl. Ukážeme dvě nedávná zlepšení: • Postavili jsme nové jazykové modely z volne dostupný schopné (šoural) verze České národní korespondence hnis a ukazují, že tyto provádět trvale lepší na texty vyráběných jak rodilými mluvčími a non-nativní studenti češtiny. • Trénovali jsme nové modely chyb na ručně s poznámkami žák korpus a ukázat, že lepší výkon než Standardní model chyba (detekce chyb) nejenom pro texty studenty ", ale také pro naše standardní hodšpatne rozpoznaných zpráv data rodilého Čecha. Pro korekci chyb se standardní model chyba překonaly non-nativní modulárně els ve 2 ze 3 testovaných datových sad. Diskutujeme důvody pro tento ne zcela intuitivní zlepprostředí. Na základě těchto poznatků a na základě analýzy chyb základě v obou domorodec a češtině frekventantů, navrhujeme směry pro další zlepšení korektor.
[ "Loganathan Ramasamy", "Alexandr Rosen", "Pavel Straňák" ]
https://www.semanticscholar.org/paper/8b394737caf81a6de812182d2715469db1110384/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Finding Nexus in the PDiT and GECCo Annotation Schemes
Hledání společných rysů v anotačních schématech PDiT a GECCo
The aim of this work is (i) to compare two frameworks for the analysis and annotation of discourse-structuring devices (DSDs) and further discourse phenomena in GECCo X PDiT and (ii) identify commonalities and/or differences between the two frameworks
Cílem práce je (i) srovnat přístupy k analýze a anotaci textových jevů v PDiT a GECCo a (ii) určit společné a rozdílné rysy mezi odpovídajícími vědeckymi paradigmaty.
[ "Ekaterina Lapshinova-Koltunski", "Anna Nedoluzhko", "Kerstin Anna Kunz", "Lucie Poláková", "Jiří Mírovský", "Pavlína Jínová" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Linguistically-augmented Perplexity-based Data Selection for Language Models
Lingivisticky vylepšený výběr dat pro jazykové modelování na základě perplexity
This paper explores the use of linguistic information for the selection of data to train language models. We depart from the state-of-the-art method in perplexity-based data selection and extend it in order to use word-level linguistic units (i.e. lemmas, named entity categories and part-of-speech tags) instead of surface forms. We then present two methods that combine the different types of linguistic knowledge as well as the surface forms (1, naıve selection of the top ranked sentences selected by each method; 2, linear interpolation of the datasets selected by the different methods). The paper presents detailed results and analysis for four languages with different levels of morphologic complexity (English, Spanish, Czech and Chinese). The interpolation-based combination outperforms the purely statistical baseline in all the scenarios, resulting in language models with lower perplexity. In relative terms the improvements are similar regardless of the language, with perplexity reductions achieved in the range 7.72% to 13.02%. In absolute terms the reduction is higher for languages with high type-token ratio (Chinese, 202.16) or rich morphology (Czech, 81.53) and lower for the remaining languages, Spanish (55.2) and English (34.43 on the same dataset as Czech and 61.90 on the same dataset as Spanish).
Tento článek se zabývá použitím lingvistiké informace pro výběr dat pro trénování jazykových modelů. Navrhovaná metoda vychází ze známých a používaných postupů, které využívají povrchových tvarů slov, a obohacuje je o informace o lemmatech, pojmenovaných entitách a slovních druzích.
[ "Antonio Toral", "Pavel Pecina", "Longyue Wang", "Josef Genabith" ]
https://www.semanticscholar.org/paper/a13ed3a2031fd0a6b663af45fc58a269b2d2e22e/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Matching Illustrative Images to “Soft News” Articles
Propojení ilustrativních obrázků s novinovými články
This work presents ongoing research on finding illustrative images to “soft news” articles, magazine-style texts where the appropriateness of an illustrative image is not judged primarily by its descriptiveness. We describe our baselines and experiments using Denoising Autoencoders and present the web-pic multimodal dataset of Czech news articles and their accompanying images. Finally, we briefly present the Safire library, a Python framework for building and analyzing complex experimental pipelines suitable for multimodal tasks.
Článek představuje probíhající výzkum zaměřený na vyhledávání ilustrativních obrázků pro články "soft news", kde vhodnost ilustrativního obrázku není primárně určena jeho popisností. Popisujeme základní řešení a experimenty používající Denoising Autoencoders.
[ "Jan Hajič, jr.", "Pavel Pecina" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
A Comparison of MT Methods for Closely Related Languages: A Case Study on Czech - Slovak and Croatian - Slovenian Language Pairs
Porovnání metod automatického překladu příbuzných jazyků - případová studie jazykových párů čeština - slovenština a chorvatština - slovinština
This paper describes an experiment comparing results of machine translation between two pairs of related Slavic languages. Two language pairs on three different translation platforms were observed in the experiment. One pair represents really very close languages (Czech and Slovak), the other pair are slightly less similar languages (Slovenian and Croatian). The comparison is performed by means of three MT systems, one for each pair representing rule-based approach, the other one representing statistical (same system for both language pairs) approach to the task. Both sets of results are manually evaluated by native speakers of the target language. The results are discussed both from the linguistic and quantitative points of view.
Článek popisuje případovou studii automatického překladu mezi příbuznými jazyky. Porovnává systémy pracující tradičními metodami s překladačem firmy Google. Porovnání je založeno na lidském hodnocení 200 náhodně vybraných vět z novinových článků.
[ "Jernej Vičič", "Vladislav Kuboň" ]
https://www.semanticscholar.org/paper/5b80e3200d5ecc400feb130281e41a3a243a9e7a/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Incremental LSTM-based Dialog State Tracker
Inkrementální sledování dialogového stavu s pomocí LSTM sítí
A dialog state tracker is an important component in modern spoken dialog systems. We present an incremental dialog state tracker, based on LSTM networks. It directly uses automatic speech recognition hypotheses to track the state. We also present the key non-standard aspects of the model that bring its performance close to the state-of-the-art and experimentally analyze their contribution: including the ASR confi- dence scores, abstracting scarcely represented values, including transcriptions in the training data, and model averaging
Sledování dialogového stavu je důležitá komponenta moderních dialogových systémů. Ukazujeme inkrementální sledovač dialogového stavu založený na LSTM sítích. Ke sledování používá přímo výsledky rozpoznávání řeči. Ukazujeme klíčová nestandardní aspekty modelu, které pomáhají dostat úspěšnost sledovače k systémům v současném stavu poznání: zahrnutí skóre rozpoznávače, abstrakce málo viděných hodnot, použití transkripce pro trénování a průměrování modelu.
[ "Lukáš Žilka", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/3e0336ff15f461bb9110652c7a919c9b2c44b7a1/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
LecTrack: Incremental Dialog State Tracking with Long Short-Term Memory Networks
LecTrack: Inkrementální odhad dialogového stavu pomocí sítí s dlouhou krátkodobou pamětí
A dialog state tracker is an important component in modern spoken dialog systems. We present the first trainable incremental dialog state tracker that directly uses automatic speech recognition hypotheses to track the state. It is based on a long short-term memory recurrent neural network, and it is fully trainable from annotated data. The tracker achieves promissing performance on the Method and Requested tracking sub-tasks in DSTC2.
Sledování dialogového stavu je důležitou součástí moderních řečových dialogových systémů. Navrhujeme první inkrementální trénovatelný sledovač dialogového stavu, který používá přímo výstup z rozpoznávače řeči ke sledování stavu. Je založen na sítích s dlouhou krátkodobou pamětí a je plně trénovatelný z anotovaných dat. Dosahujeme slibných výsledků na sledování pod-úkolů Method a Requested v DSTC2.
[ "Lukáš Žilka", "Filip Jurčíček" ]
https://www.semanticscholar.org/paper/6d10e4714d83335fd9c0ecf7b006c8d1f17f743d/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Towards Cross-language Application of Dependency Grammar
Směřování ke křížovému použití gramatiky závislosti
This paper discusses the adaptation of the Stanford typed dependency model (de Marneffe and Manning 2008), initially designed for English, to the requirements of typologically different languages from the viewpoint of practical parsing. We argue for a framework of functional dependency grammar that is based on the idea of parallelism between syntax and semantics. There is a twofold challenge: (1) specifying the annotation scheme in order to deal with the morphological and syntactic peculiarities of each language and (2) maintaining crosslinguistically consistent annotations to ensure homogenous analysis for similar linguistic phenomena. We applied a number of modifications to the original Stanford scheme in an attempt to capture the language-specific grammatical features present in heterogeneous CoNLL-encoded data sets for German, Dutch, French, Spanish, Brazilian Portuguese, Russian, Polish, Indonesian, and Traditional Chinese. From a multilingual perspective, we discuss features such as subject and object verb complements, comparative phrases, expletives, reduplication, copula elision, clitics and adpositions.
Tento příspěvek pojednává o adaptaci modelu závislého typu Stanfordova typu (de Marneffe a Manning 2008) původně určeného pro angličtinu na požadavky typologicky odlišných jazyků z hlediska praktického rozboru. Argumentujeme pro rámec gramatiky funkční závislosti, který je založen na myšlence paralelnosti mezi syntaxí a sémantikou.
[ "Timo Järvinen", "Elisabeth Bertol", "Septina Dian Larasati", "Monica-Mihaela Rizea", "Maria Ruiz Santabalbina", "Milan Souček" ]
https://www.semanticscholar.org/paper/cfe2799f41686c2290c2dd17d16a87b99c55a22f/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
RExtractor: a Robust Information Extractor
RExtractor: robustní informační extraktor
The RExtractor system is an information extractor that processes input documents by natural language processing tools and consequently queries the parsed sentences to extract a knowledge base of entities and their relations. The extraction queries are designed manually using a tool that enables natural graphical representation of queries over dependency trees. A workflow of the system is designed to be language and domain independent. We demonstrate RExtractor on Czech and English legal documents.
RExtractor je systém pro extrakci informací. Vstupní dokumenty jsou zpracovány NLP nástroji. Extrakce následně probíhá pomocí dotazů nad závislostními stromy. Výsledkem je znalostní báze dokumentu, definována jako množina entit a vztahů mezi nima. Dotazy nad stromy jsou definovány manuálně. Architektura systému je navržena doménově a jazykově nezávisle. Systém demonstrujeme na českých a anglických právních dokumentech.
[ "Vincent Kríž" ]
https://www.semanticscholar.org/paper/6c524c3fcdacd156a1cd074700843cd2bf04c712/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Feature Extraction for Native Language Identification Using Language Modeling
Extrakce rysů založených na jazykovém modelování pro úlohu identifikace mateřského jazyka
This paper reports on the task of Native Language Identification (NLI). We developed a machine learning system to identify the native language of authors of English texts written by non-native English speakers. Our system is based on the language modeling approach and employs cross-entropy scores as features for supervised learning, which leads to a significantly reduced feature space. Our method uses the SVM learner and achieves the accuracy of 82.4 % with only 55 features. We compare our results with the previous similar work by Tetreault et al. (2012) and analyze more details about the use of language modeling for NLI. We experiment with the TOEFL11 corpus (Blanchard et al., 2013) and provide an exact comparison with results achieved in the First Shared Task in NLI (Tetreault et al., 2013).
Prezentujeme výsledky experimentů na úloze identifikace mateřského jazyka. Vytvořili jsme systém pro automatickou identifikaci mateřského jazyka autora anglicky psaného textu. Systém je založen na jazykovém modelování a využíva rysy spočítané na základe křížové entropie.
[ "Vincent Kríž", "Martin Holub", "Pavel Pecina" ]
https://www.semanticscholar.org/paper/03abf83573fb6c9700e27bf812c6872f7c91b366/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
RExtractor: a Robust Information Extractor
RExtractor: robustní informační extraktor
The RExtractor system is an information extractor that processes input documents by natural language processing tools and consequently queries the parsed sentences to extract a knowledge base of entities and their relations. The extraction queries are designed manually using a tool that enables natural graphical representation of queries over dependency trees. A workflow of the system is designed to be language and domain independent. We demonstrate RExtractor on Czech and English legal documents.
RExtractor je systém pro extrakci informací. Vstupní dokumenty jsou zpracovány NLP nástroji. Extrakce následně probíhá pomocí dotazů nad závislostními stromy. Výsledkem je znalostní báze dokumentu, definována jako množina entit a vztahů mezi nima. Dotazy nad stromy jsou definovány manuálně. Architektura systému je navržena doménově a jazykově nezávisle. Systém demonstrujeme na českých a anglických právních dokumentech.
[ "Vincent Kríž", "Barbora Hladká" ]
https://www.semanticscholar.org/paper/6c524c3fcdacd156a1cd074700843cd2bf04c712/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
Czech Legal Text Treebank 1.0
Czech Legal Text Treebank 1.0
The Czech Legal Text Treebank (CLTT) is a collection of 1133 manually annotated dependency trees. CLTT consists of two legal documents: The Accounting Act (563/1991 Coll., as amended) and Decree on Double-entry Accounting for undertakers (500/2002 Coll., as amended).
Korpus Czech Legal Text Treebank (CLTT) je kolekcí 1 133 ručně syntakticky anotovaných vět. Věty pochází ze Zákona o účetnictví (563/1991 Sb.) a z Vyhlášky o účetnictví (500/2002 Sb.).
[ "Vincent Kríž", "Barbora Hladká", "Zdeňka Urešová" ]
https://www.semanticscholar.org/paper/a64dcda9b1a7ad699e42dcd186f8e7d8c2113e5a/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
LINDAT/CLARIN repository for language data and tools (DSpace@UFAL)
Repozitář pro jazyková data a nástroje LINDAT/CLARIN (DSpace@ÚFAL)
The presentation introduces the LINDAT/CLARIN repository and some of the modifications to DSpace - the platform the repository is based on.
Prezentace představuje repozitář LINDAT/CLARIN a některé změny platformy DSpace, na které je repozitář založený.
[ "Ondřej Košarko" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Obituary - Jane J. Robinson
Nekrolog - Jane J. Robinson
Obituary article evaluating the scientific contribution of J.J.Robinson, an American computational linguist, one of the founders of the field.
Nekrolog obsahující zhodnocení vědeckého přínosu jedné ze zakladatelek počítačové lingvistiky, americké badatelky prof. J. J. Robinson.
[ "Barbara J. Grosz", "Eva Hajičová", "Aravind K. Joshi" ]
https://www.semanticscholar.org/paper/20eb50d84a9242ccaaf64c4ac9a316d0068b20f0/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
CLARIN Concept Registry: the new semantic registry replacing ISOcat
CLARIN Concept Registry: nový sémantický registr, který nahrazuje ISOcat
We introduce the CLARIN Concept Registry, a new on-line registry of semantic concepts that can be used in (and referenced from) CLARIN projects. This new registry replaces (for CLARIN purposes) ISOcat. We discuss various issues that made ISOcat less useful, and propose solutions to those issues in CCR.
Představujeme CLARIN Concept Registry, nový registr sémantických konceptů, který je dostupný on-line a jeho smyslem je sloužit jako definice konceptů, na které se odkazují projekty CLARIN. Tento nový registr nahrazuje (pro účely CLARINu) ISOcat. Probíráme různé problémy, které komplikovaly používání ISOcatu, a navrhujeme řešení těchto problémů v CCR.
[ "Ineke Schuurman", "Menzo Windhouwer", "Oddrun Ohren", "Daniel Zeman" ]
https://www.semanticscholar.org/paper/8b35185db2b60318579c741f18f19eb1ec0fa448/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Evaluating Machine Translation Quality Using Short Segments Annotations
Vyhodnocování kvality strojového překladu anotací krátkých úseků
We propose a manual evaluation method for machine translation (MT), in which annotators rank only translations of short segments instead of whole sentences. This results in an easier and more efficient annotation. We have conducted an annotation experiment and evaluated a set of MT systems using this method. The obtained results are very close to the official WMT14 evaluation results. We also use the collected database of annotations to automatically evaluate new, unseen systems and to tune parameters of a statistical machine translation system. The evaluation of unseen systems, however, does not work and we analyze the reasons.
Studujeme metodu ručního hodnocení strojového překladu, v níž anotátoři uspořádávají podle kvality jen krátké úseky místo celých vět. Anotace je tak snazší a rychlejší.
[ "Matouš Macháček", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/7f4f85d80a87a4c06c6b1c0d42de1368f92a665f/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Results of the WMT15 Tuning Shared Task
Výsledky soutěže v ladění systémů strojového překladu WMT15
This paper presents the results of the WMT15 Tuning Shared Task. We provided the participants of this task with a complete machine translation system and asked them to tune its internal parameters (feature weights). The tuned systems were used to translate the test set and the outputs were manually ranked for translation quality. We received 4 submissions in the English-Czech and 6 in the Czech-English translation direction. In addition, we ran 3 baseline setups, tuning the parameters with standard optimizers for BLEU score.
Článek představuje výsledky soutěže v ladění systémů strojového překladu, WMT15 Tuning Shared Task. Účastníkům jsme poskytli úplný překladový systém a jejich úkolem bylo nastavit jeho interní váhy. Výsledné překlady byly porovnány v ručním hodnocení.
[ "Miloš Stanojević", "Amir Kamran", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/640707a70deb36dc41a22b5409246b5aad86438e/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Results of the WMT15 Metrics Shared Task
Výsledky soutěže v automatickém hodnocení strojového překladu WMT15
This paper presents the results of the WMT15 Metrics Shared Task. We asked participants of this task to score the outputs of the MT systems involved in the WMT15 Shared Translation Task. We collected scores of 46 metrics from 11 research groups. In addition to that, we computed scores of 7 standard metrics (BLEU, SentBLEU, NIST, WER, PER, TER and CDER) as baselines. The collected scores were evaluated in terms of system level correlation (how well each metric's scores correlate with WMT15 official manual ranking of systems) and in terms of segment level correlation (how often a metric agrees with humans in comparing two translations of a particular sentence).
Článek představuje výsledky soutěže v automatickém hodnocení kvality strojového překladu (WMT15 Metrics Shared Task). Úkolem účastníků bylo vyhodnotit kvalitu překladových systémů, které se zúčastnily překladové úlohy WMT15. Získaných 46 metrik od 11 týmů jsme pak společně se 7 základními metrikami (BLEU, SentBLEU, NIST, WER, PER, TER a CDER) porovnali z hlediska korelace s lidským hodnocením pro celý testset i pro jednotlivé věty.
[ "Miloš Stanojević", "Amir Kamran", "Philipp Koehn", "Ondřej Bojar" ]
https://www.semanticscholar.org/paper/2da338d8972e473df62a566290c9de95a52209e5/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Zero Alignment of Verb Arguments in a Parallel Treebank
Nepropojené slovesné argumenty v paralelním korpusu
This paper analyses several points of interlingual dependency mismatch on the material of a parallel Czech-English dependency treebank. Particularly, the points of alignment mismatch between the valency frame arguments of the corresponding verbs are observed and described. The attention is drawn to the question whether such mismatches stem from the inherent semantic properties of the individual languages, or from the character of the used linguistic theory. Comments are made on the possible shifts in meaning. The authors use the findings to make predictions about possible machine translation implementation of the data.
Článek popisuje několik případů mezijazykových rozdílů závislosti na materiálu paralelního češko-anglického závislostního korpusu. Pozornost je soustředěna zejména na místa, kde dochází k neshodám v propojení slovesných argumentů. Článek se věnuje otázce, zda takové neshody pramení ze sémantických vlastností jednotlivých jazyků, nebo z charakteru použité lingvistické teorie. Autoři se zamýšlejí nad případným využitím získaných poznatků pro strojový překlad.
[ "Jana Šindlerová", "Eva Fučíková", "Zdeňka Urešová" ]
https://www.semanticscholar.org/paper/104bca087863214853e935f1b59e101e354d23cc/
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Discourse and Coherence. From the Sentence Structure to Relations in Text
Diskurz a koherence. Od struktury věty k vztahům v textu
In this monograph we present the results of our research on the interplay of intra-sentential relations such as deep syntactic relations and information structure of the sentence and the inter-sentential relations such as discourse relations and coreferential and other associative links.
V monografii je představen výzkum vztahů působících společně ve výstavbě textu (syntaktická výstavba, aktuální členění, sémantické diskurzní vztahy v užším smyslu, koreference a asociativní anafora).
[ "Šárka Zikánová", "Eva Hajičová", "Barbora Hladká", "Pavlína Jínová", "Jiří Mírovský", "Anna Nedoluzhko", "Lucie Poláková", "Kateřina Rysová", "Magdaléna Rysová", "Jan Václ" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
Annotation of the text structure in the Prague Dependency Treebank
Zachycení výstavby textu v Pražském závislostním korpusu
Information about annotation of discourse relations, coreference and bridging anaphora in the Prague Dependency Treebank
Seznámení s anotací diskurzních vztahů a koreference a asociační anafory v Pražském závislostním korpusu
[ "Šárka Zikánová", "Lucie Poláková", "Pavlína Jínová", "Anna Nedoluzhko", "Magdaléna Rysová", "Jiří Mírovský", "Eva Hajičová" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Retrieving information about medical symptoms
Vyhledávání informací o zdravotních symptomech
This paper details methods, results and analysis of the CLEF 2015 eHealth Evaluation Lab, Task 2. This task investigates the effectiveness of web search engines in providing access to medical information with the aim of fostering advances in the development of these technologies
Článek popisuje metody a výsledky CLEF 2015 eHealth Evaluation Lab, Task 2, který se zaměřuje na efektivitu vyhledávání medicínských informací na webu.
[ "João Palotti", "Guido Zuccon", "Lorraine Goeuriot", "Liadh Kelly", "Allan Hanbury", "Gareth J.F. Jones", "Mihai Lupu", "Pavel Pecina" ]
https://www.semanticscholar.org/paper/59ee8fddcc143d29dc4cf569320a0ee6b3f7eef2/
null
null
null
null
null
null
null
null
null
null
null
null
cs
2015
Discourse Connectives in Czech (From Centre to Periphery)
Diskurzní konektory v češtině (Od centra k periferii)
The thesis focuses on description and analysis of discourse connectives in Czech in broader sense, i.e. by which language means it is possible to express sense relation within a text. The thesis is not limited to any parts of speech (like conjunctions or structuring particles) but it tries to find and describe all language means in Czech with the ability to connect two pieces or units of a text into one coherent complex. The thesis investigates discourse connectives in Czech with respect to the so called secondary connectives (i.e. mainly multiword phrases like "to je důvod, proč" – "that is the reason why"; "kvůli těmto skutečnostem" – "due to these facts" etc., in opposition to primary connectives like "však" – "however", "nebo" – "or", "a" – "and", "ale" – "but", "proto" – "therefore" etc.).
V práci se zabýváme popisem a analýzou diskurzních (tj. textových) konektorů v češtině v širším smyslu, tedy tím, jakými jazykovými prostředky je možné vyjadřovat v textu diskurzní vztahy. Výzkum přitom neomezujeme na předem stanovenou skupinu výrazů (danou například příslušností k určitým slovním druhům, jako jsou spojky či strukturující částice), ale snažíme se nalézt a obecně popsat všechny jazykové prostředky v češtině, které mají schopnost spojovat jednotlivé úseky textu v jeden koherentní celek. Zaměřujeme se především na méně probádané víceslovné konektivní struktury typu "to je důvod, proč"; "kvůli těmto skutečnostem"; "z těchto důvodů" atd., pro které užíváme označení sekundární konektory (za primární konektory považujeme především konektivní synsémantika typu "však", "nebo", "a", "ale", "proto" apod.).
[ "Magdaléna Rysová" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
On Definition of Discourse Connectives: Primary vs. Secondary Connectives (Based on a Corpus Probe)
K definici diskurzních konektorů: primární a sekundární konektory (korpusová studie)
The aim of the poster is to contribute to the general discussion on discourse connectives, especially on their definition and principles we may hold as boundaries surrounding this class of expressions.
Cílem posteru je přispět do diskuze o diskurzních konektorech, zejména o jejich definici a o kritériích, podle kterých je možné vymezit jejich hranice.
[ "Magdaléna Rysová", "Kateřina Rysová" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
en
2015
Secondary Connectives in the Prague Dependency Treebank
Sekundární konektory v Pražském závislostním korpusu
The paper introduces a new annotation of discourse relations in the Prague Dependency Treebank (PDT), i.e. the annotation of the so called secondary connectives (mainly multiword phrases like "the condition is", "that is the reason why", "to conclude", "this means" etc.). Firstly, the paper concentrates on theoretical introduction of these expressions (mainly with respect to primary connectives like "and", "but", "or", "too" etc.) and tries to contribute to the description and definition of discourse connectives in general (both primary and secondary). Secondly, the paper demonstrates possibilities of annotations of secondary connectives in large corpora (like PDT). The paper describes general annotation principles for secondary connectives used in PDT for Czech and compares the results of this annotation with annotation of primary connectives in PDT. In this respect, the main aim of the paper is to introduce a new type of discourse annotation that could be adopted also by other languages.
Článek představuje novou anotaci diskurzních vztahů v Pražském závislostním korpusu (PDT) – anotaci tzv. sekundárních konektorů (většinou víceslovných frází jako "podmínkou je", "to je důvod, proč", "na závěr", "to znamená" atd.). Nejprve se článek soustředí na na teoretické vymezení těchto výrazů (hlavně s ohledem na tzv. primární konektory jako "a", "ale", "nebo", "také" atd.) a pak se zaměřuje na popis a definici diskurzních konektorů jako takových (primárních i sekundárních). Poté článek představuje možnosti anotací sekundárních konektorů ve velkých korpusech (jako je PDT). Příspěvek popisuje obecné anotační principy pro sekundární konektory použité v PDT pro češtinu a srovnává výsledky této anotace s výsledky anotací primárních konektorů v PDT. Hlavním cílem příspěvku je představit nový typ anotace diskurzu, která by mohla být využita i pro další jazyky.
[ "Magdaléna Rysová", "Kateřina Rysová" ]
https://www.semanticscholar.org/paper/4dade01ea3fcb35e3ec96714ded2048e4f9ab473/
null
null
null
null
null
null
null
null
null
null
null
null