sentence1
stringlengths 16
446
| sentence2
stringlengths 14
436
|
|---|---|
socher et al assign a vector and a matrix to each word for the purpose of semantic composition , and build recursive neural network along constituency tree . | socher et al build rnn on constituency trees of sentences , and apply the model to relation recognition task . |
in this paper , we learn the semantic contribution of characters to a word by exploiting the similarity between a word and its component characters . | in this paper , we exploit the internal structure in chinese words by learning the semantic contribution of internal characters to the word . |
sentence compression is the task of compressing long , verbose sentences . | sentence compression can be seen as sentence-level summarization . |
event extraction is a particularly challenging information extraction task , which intends to identify and classify event triggers and arguments from raw text . | thus , event extraction is a difficult task and requires substantial training data . |
we use the average glove embedding as the sentence embedding . | for the word-embedding based classifier , we use the glove pre-trained word embeddings . |
accuracy , we then extract humor anchors in sentences via a simple and effective method . | furthermore , we develop a simple and effective method to extract anchors that enable humor in a sentence . |
we represent each word by a vector with length 300 . | the dimension of glove word vectors is set as 300 . |
system incorporating a phrase-based error model significantly outperforms its baseline systems . | results show that the system using the phrase-based error model outperforms significantly its baseline systems . |
dhingra et al proposed an end-to-end differentiable kb-infobot for efficient information access . | li et al and dhingra et al also proposed end-to-end task-oriented dialog models that can be trained with hybrid supervised learning and rl . |
we propose a joint model for answer sentence ranking and answer extraction . | we present a joint model for the important qa tasks of answer sentence ranking and answer extraction . |
we use stanford named entity recognizer 7 to extract named entities from the texts . | we use the stanford named entity recognizer for this purpose . |
semantic role labeling ( srl ) is the process of producing such a markup . | semantic role labeling ( srl ) is the task of identifying the predicate-argument structure of a sentence . |
in fact , the rule-based system of raghunathan et al exhibited the top score in the recent conll evaluation . | this capability is very desirable as shown by the success of the rule-based deterministic approach of raghunathan et al in the conll shared task 2011 . |
word alignment is the task of identifying word correspondences between parallel sentence pairs . | word alignment is a key component in most statistical machine translation systems . |
one solution is to consider only the normal-form derivation , which is the route taken in hockenmaier and steedman . | we use the baseline model of hockenmaier and steedman , which is a simple generative model that is equivalent to an unlexicalized pcfg . |
we participated only in the task 2a in which the gold standard disorder mentions were given . | we participated only in the disorder attribute detection task 2a . |
that show substantial improvements in spearman correlation scores over the baseline models provided by task 1 organizers , ( ranging from 0 . 03 to 0 . 23 ) . | the performance of the system for all subtasks in both languages shows substantial improvements in spearman correlation scores over the baseline models provided by task 1 organizers , ranging from 0.03 to 0.23 . |
the word embeddings are initialized with 100-dimensions vectors pre-trained by the cbow model . | we initialize the embedding weights by the pre-trained word embeddings with 200 dimensional vectors . |
on several data conditions , we show that our method outperforms the baseline and results in up to 8 . 5 % improvement . | on several data conditions , we show that our method outperforms the baseline and results in up to 8.5 % improvement in the f 1 -score . |
we think the natural language and speech processing technology will be useful for the efficient production of tv programs with closed captions . | we propose natural language and speech processing techniques should be used for efficient closed caption production of tv programs . |
we used the srilm toolkit to build unpruned 5-gram models using interpolated modified kneser-ney smoothing . | for all experiments , we used a 4-gram language model with modified kneser-ney smoothing which was trained with the srilm toolkit . |
coreference resolution is the task of partitioning a set of mentions ( i.e . person , organization and location ) into entities . | coreference resolution is a well known clustering task in natural language processing . |
we also use glove vectors to initialize the word embedding matrix in the caption embedding module . | we initialize the embedding weights by the pre-trained word embeddings with 200 dimensional vectors . |
distributed word representations induced through deep neural networks have been shown to be useful in several natural language processing applications . | word embeddings have proved useful in downstream nlp tasks such as part of speech tagging , named entity recognition , and machine translation . |
a 5-gram language model was built using srilm on the target side of the corresponding training corpus . | the language model is a 3-gram language model trained using the srilm toolkit on the english side of the training data . |
we compared sn models with two different pre-trained word embeddings , using either word2vec or fasttext . | we also used word2vec to generate dense word vectors for all word types in our learning corpus . |
misra et al use a latent dirichlet allocation topic model to find coherent segment boundaries . | brody and lapata extend the latent dirichlet allocation model to combine evidence from different types of contexts . |
in this paper , we focus on class-based models of selectional preferences . | in this paper , we evaluated five models for the acquisition of selectional preferences . |
pang et al observed that the top 2633 unigrams are better features than unigrams or adjectives for sentiment classification of a document . | pang et al proved that unigrams and bigrams , adjectives and part of speech tags are important features for a machine learning based sentiment classifier . |
however , only a few techniques to learn finite-state transducers for machine translation purposes can be found . | nowadays , only a few techniques exist for inferring finite-state transducers . |
relation extraction is the task of detecting and classifying relationships between two entities from text . | relation extraction ( re ) is the task of extracting instances of semantic relations between entities in unstructured data such as natural language text . |
we show that this method achieves state of the art performance . | we have shown that state of the art performance can be achieved by using this approach . |
the standard minimum error rate training algorithm was used for tuning . | parameters were tuned using minimum error rate training . |
coreference resolution is the task of partitioning a set of entity mentions in a text , where each partition corresponds to some entity in an underlying discourse model . | coreference resolution is a well known clustering task in natural language processing . |
stephens et al propose 17 classes targeted to relations between genes . | stephens et al propose 17 very specific classes targeting relations between genes . |
however , dependency parsing , which is a popular choice for japanese , can incorporate only shallow syntactic information , i.e. , pos tags , compared with the richer syntactic phrasal categories in constituency parsing . | dependency parsing is the task of building dependency links between words in a sentence , which has recently gained a wide interest in the natural language processing community . |
in the work of wang et al , a variant of attention-based lstm was proposed . | wang et al utilized attention-based lstm , which takes into account aspect information during attention . |
from the perspective of online language comprehension , processing difficulty is quantified by surprisal . | pcfg surprisal is a measure of incremental hierarchic syntactic processing . |
the task of automatically assigning predefined meanings to words in contexts , known as word sense disambiguation , is a fundamental task in computational lexical semantics . | word sense disambiguation , the task of automatically assigning predefined meanings to words occurring in context , is a fundamental task in computational lexical semantics . |
we use phrase-based and hierarchical mt systems as implemented by koehn et al for our experiments . | we have implemented a hierarchical phrase-based smt model similar to chiang . |
we use srilm for training the 5-gram language model with interpolated modified kneser-ney discounting . | we estimated 5-gram language models using the sri toolkit with modified kneser-ney smoothing . |
pcfg surprisal is a measure of incremental hierarchic syntactic processing . | from the perspective of online language comprehension , processing difficulty is quantified by surprisal . |
luong et al segment words using morfessor , and use recursive neural networks to build word embeddings from morph embeddings . | luong et al created a hierarchical language model that uses rnn to combine morphemes of a word to obtain a word representation . |
rewrite rules are used in many areas of natural language and speech processing , including syntax . | context-dependent rewrite rules are used in many areas of natural language and speech processing . |
keyphrases also offers a programming framework for developing new extraction . | alchemyapi 8 offers a web service for keyword extraction . |
we demonstrate superagent as an add-on extension to mainstream web browsers such as microsoft edge and google chrome . | we demonstrate superagent as an add-on extension to mainstream web browsers and show its usefulness to user ’ s online shopping experience . |
yao et al attempted to improve the specificity with the reinforcement learning framework by using the averaged idf score of the words in the response as a reward . | yao et al diversified the response by a loss function in which words with high inverse document frequency values are preferred . |
semantic role labeling ( srl ) is a task of analyzing predicate-argument structures in texts . | semantic role labeling ( srl ) is defined as the task to recognize arguments for a given predicate and assign semantic role labels to them . |
automatic detection of semantic roles has received a lot of attention lately . | copious work has been done lately on semantic roles . |
in natural language , a word often assumes different meanings , and the task of determining the correct meaning , or sense , of a word in different contexts is known as word sense disambiguation ( wsd ) . | word sense disambiguation ( wsd ) is a problem of finding the relevant clues in a surrounding context . |
we trained a specific language model using srilm from each of these corpora in order to estimate n-gram log-probabilities . | for both languages , we used the srilm toolkit to train a 5-gram language model using all monolingual data provided . |
we apply a pretrained glove word embedding on . | we use pre-trained embeddings from glove . |
parameter optimisation is done by mini-batch stochastic gradient descent where back-propagation is performed using adadelta update rule . | training is done using stochastic gradient descent over mini-batches with the adadelta update rule . |
coreference resolution is a key task in natural language processing ( cite-p-13-1-8 ) aiming to detect the referential expressions ( mentions ) in a text that point to the same entity . | coreference resolution is the task of clustering a set of mentions in the text such that all mentions in the same cluster refer to the same entity . |
we preprocessed all the corpora used with scripts from the moses toolkit . | we tokenized , cleaned , and truecased our data using the standard tools from the moses toolkit . |
soricut and echihabi explore pseudo-references and document-aware features for document-level ranking , using bleu as quality label . | soricut and echihabi propose documentlevel features to predict document-level quality for ranking purposes , having bleu as quality label . |
annotation was conducted on a modified version of the brat web-based annotation tool . | the annotation was performed manually using the brat annotation tool . |
negation is a grammatical category which comprises various kinds of devices to reverse the truth value of a proposition ( cite-p-18-3-8 ) . | although negation is a very relevant and complex semantic aspect of language , current proposals to annotate meaning either dismiss negation or only treat it in a partial manner . |
experimental results show substantial improvements of the acm in comparison with classical cluster models and word n-gram models . | results show approximately 6-10 % cer reduction of the acms in comparison with the word trigram models , even when the acms are slightly smaller . |
we use the stanford parser to extract a set of dependencies from each comment . | we use the stanford parser to generate a dg for each sentence . |
in ( 2 ) , however , it seems clear from context that we are dealing with an unpleasant person for whom laugh . | in ( 2 ) , however , it seems clear from context that we are dealing with an unpleasant person for whom laugh entails bitter laugh . |
semantic role labeling ( srl ) is defined as the task to recognize arguments for a given predicate and assign semantic role labels to them . | semantic role labeling ( srl ) is the task of automatic recognition of individual predicates together with their major roles ( e.g . frame elements ) as they are grammatically realized in input sentences . |
secondly , a knowledge-based criterion is used to supervise the hierarchical splitting of these semantic-related tags . | by introducing a knowledge-based criterion , these new tags are decided whether or not to split into subcategories from a semantic perspective . |
curran and moens have demonstrated that dramatically increasing the volume of raw input text used to extract context information significantly improves the quality of extracted synonyms . | curran and moens found that dramatically increasing the volume of raw input data for distributional similarity tasks increases the accuracy of synonyms extracted . |
to calculate language model features , we train traditional n-gram language models with ngram lengths of four and five using the srilm toolkit . | we use srilm toolkit to train a trigram language model with modified kneser-ney smoothing on the target side of training corpus . |
we used a 5-gram language model trained on 126 million words of the xinhua section of the english gigaword corpus , estimated with srilm . | we used the srilm toolkit to create 5-gram language models with interpolated modified kneser-ney discounting . |
conjecture and empirically show that entailment graphs exhibit a ¡° tree-like ¡± property , i . e . , that they can be reduced into a structure similar to a directed forest . | we first identify that entailment graphs exhibit a ¡°tree-like¡± property and are very similar to a novel type of graph termed forest-reducible graph . |
further , we apply a 4-gram language model trained with the srilm toolkit on the target side of the training corpus . | we employ srilm toolkit to linearly interpolate the target side of the training corpus with the wmt english corpus , optimizing towards the mt tuning set . |
sentiment analysis is the task of automatically identifying the valence or polarity of a piece of text . | sentiment analysis is the task of identifying the polarity ( positive , negative or neutral ) of review . |
modified kneser-ney trigram models are trained using srilm on the chinese portion of the training data . | the language model is a 3-gram language model trained using the srilm toolkit on the english side of the training data . |
in this paper we present l obby b ack , a system to reconstruct the “ dark corpora ” that is comprised of model . | in this paper we present l obby b ack , a system that reverse engineers model legislation from observed text . |
automatically acquired lexicons with subcategorization information have already proved accurate and useful enough for some purposes . | at present , automatically acquired verb lexicons with scf information have already proved accurate and useful enough for some nlp purposes ( cite-p-8-3-5 , cite-p-8-3-3 ) . |
a tri-gram language model is estimated using the srilm toolkit . | a 4-grams language model is trained by the srilm toolkit . |
we introduce a new clustering method called hierarchical graph factorization clustering ( hgfc ) . | we introduce then a new method called hierarchical graph factorization clustering ( hgfc ) ( cite-p-17-5-8 ) . |
in machine translation and text summarization , results are automatically evaluated based on sentence comparison . | in text summarization and machine translation , summaries comparison based on sentence similarity has been applied for automatic evaluation . |
we implemented the different aes models using scikit-learn . | for all classifiers , we used the scikit-learn implementation . |
a 4-gram language model was trained on the target side of the parallel data using the srilm toolkit . | we used a 5-gram language model with modified kneser-ney smoothing , built with the srilm toolkit . |
word sense disambiguation ( wsd ) is the task of automatically determining the correct sense for a target word given the context in which it occurs . | word sense disambiguation ( wsd ) is a widely studied task in natural language processing : given a word and its context , assign the correct sense of the word based on a predefined sense inventory ( cite-p-15-3-4 ) . |
relation extraction ( re ) is a task of identifying typed relations between known entity mentions in a sentence . | relation extraction is the task of finding relational facts in unstructured text and putting them into a structured ( tabularized ) knowledge base . |
we convert the question into a sequence of learned word embeddings by looking up the pre-trained vectors , such as glove . | for the classification task , we use pre-trained glove embedding vectors as lexical features . |
we used the srilm toolkit to create 5-gram language models with interpolated modified kneser-ney discounting . | we used the sri language modeling toolkit to calculate the log probability and two measures of perplexity . |
we have created the first publicly-available corpus of gold standard negative deceptive opinion spam , containing 400 reviews of 20 chicago hotels , which we have used to compare the deception detection capabilities of untrained human judges . | following the framework of cite-p-12-1-12 , we use amazon ’ s mechanical turk service to produce the first publicly available 1 dataset of negative deceptive opinion spam , containing 400 gold standard deceptive negative reviews of 20 popular chicago hotels . |
in this paper , we use the term uncertain information . | in this paper , we develop attention mechanisms for uncertainty detection . |
as is shown in , the japanese orthography is highly irregular , which contributes to a substantial number of out-of-vocabulary words in the machine translation output . | in addition , the highly irregular japanese orthography as is analyzed in poses a challenge for machine translation tasks . |
our approach to this subtask is based on the sieves proposed by lee et al . | most of these sieves are relaxed versions of the ones proposed by lee et al . |
crowdsourcing is the use of the mass collaboration of internet passersby for large enterprises on the world wide web such as wikipedia and survey companies . | crowdsourcing is a scalable and inexpensive data collection method , but collecting high quality data efficiently requires thoughtful orchestration of crowdsourcing jobs . |
krishnakumaran and zhu use the isa relation in wordnet for metaphor recognition . | krishnakumaran and zhu use wordnet knowledge to differentiate between metaphors and literal usage . |
as mentioned earlier , our approach was motivated by karttunen ' s implementation . | as mentioned earlier , our approach was motivated by karttunen 's implementation as described in karttunen 1984 . |
a 4-gram language model is trained on the monolingual data by srilm toolkit . | the language model is trained with the sri lm toolkit , on all the available french data without the ted data . |
in addition , we automatically rescale models so that they have physically plausible sizes and orient them so that they have a consistent up and front direction . | in addition , we assume the models have been scaled to physically plausible sizes and oriented with consistent up and front direction . |
in all cases , we used the implementations from the scikitlearn machine learning library . | for the feature-based system we used logistic regression classifier from the scikit-learn library . |
relation extraction ( re ) is the task of recognizing the assertion of a particular relationship between two or more entities in text . | relation extraction is the task of finding semantic relations between two entities from text . |
our model by construction is similar to approach based on the ising spin model described in . | under this setting , we compare our method to the spin model described in . |
ji et al introduced an extra latent variable to a hierarchical rnn model to represent discourse relation . | ji et al proposed a latent variable rnn for modeling discourse relations between sentences . |
events and entities is highly contextually dependent . | the interpretation of event descriptions is highly contextually dependent . |
we set all feature weights by optimizing bleu directly using minimum error rate training on the tuning part of the development set . | we used minimum error rate training to tune the feature weights for maximum bleu on the development set . |
we used a 5-gram language model trained on 126 million words of the xinhua section of the english gigaword corpus , estimated with srilm . | we trained a 4-gram language model on the xinhua portion of gigaword corpus using the sri language modeling toolkit with modified kneser-ney smoothing . |
given the model parameters and a sentence math-w-2-16-0-10 , determine the most probable translation of math-w-2-16-0-18 . | given the parameters of ibm model 3 , and a sentence pair math-w-5-1-0-21 , compute the probability math-w-5-1-0-30 . |
here we investigate the benefits of displaying the discourse structure information . | in this paper we explore the utility of the navigation map , a graphical representation of the discourse structure . |
for full text , in this paper , we introduce a large corpus of chinese short text summarization dataset constructed from the chinese microblogging website sina weibo , which is released to the public . | in this paper , we take one step back and focus on constructing lcsts , the large-scale chinese short text summarization dataset by utilizing the naturally annotated web resources on sina weibo . |