HHousen/ParaSCI · Datasets at Hugging Face

sentence1 stringlengths 16 446	sentence2 stringlengths 14 436
we use stanford part-of-speech tagger to automatically detect nouns from text .	we use stanford log-linear partof-speech tagger to produce pos tags for the english side .
as a baseline system for our experiments we use the syntax-based component of the moses toolkit .	our implementation of the segment-based imt protocol is based on the moses toolkit .
we use latent dirichlet allocation , or lda , to obtain a topic distribution over conversations .	an effective strategy to cluster words into topics , is latent dirichlet allocation .
in this paper , we present an unsupervised dynamic bayesian model that allows us to model stylistic style accommodation .	in this paper we presented an unsupervised dynamic bayesian modeling approach to modeling speech style accommodation in face-to-face interactions .
coreference resolution is the process of linking together multiple referring expressions of a given entity in the world .	coreference resolution is the next step on the way towards discourse understanding .
our evaluation on two wordnet-derived taxonomies shows that the learned taxonomies capture a higher number of correct taxonomic relations compared to those produced by traditional distributional similarity approaches .	we evaluate our method on two wordnetderived subtaxonomies and show that our method leads to the development of concept hierarchies that capture a higher number of correct taxonomic relations in comparison to those generated by current distributional similarity approaches .
for training our system classifier , we have used scikit-learn .	we implemented the different aes models using scikit-learn .
we trained a trigram language model on the chinese side , with the srilm toolkit , using the modified kneser-ney smoothing option .	we trained a 4-gram language model on this data with kneser-ney discounting using srilm .
the word embeddings are pre-trained by skip-gram .	all word vectors are trained on the skipgram architecture .
our dataset and parser can be found at http : / / www .	the dataset and parser can be found at http : //www .
parsing is a computationally intensive task due to the combinatorial explosion seen in chart parsing algorithms that explore possible parse trees .	parsing is the process of mapping sentences to their syntactic representations .
through the method , various kinds of collocations induced by key strings are retrieved .	through the method , various range of collocations which are frequently used in a specific domain are retrieved automatically .
mcclosky et al use self-training in combination with a pcfg parser and reranking .	mcclosky et al , 2006 , presents a successful instance of parsing with self-training by using a re-ranker .
our model is a first order linear chain conditional random field .	our model is a structured conditional random field .
rating scales are common in psychology and related fields , such studies hardly exist in nlp , and so .	studies assessing rating scales are very common in psychology and related fields , but are rare in nlp .
we used data from the conll-x shared task on multilingual dependency parsing .	for monolingual treebank data we relied on the conll-x and conll-2007 shared tasks on dependency parsing .
to train the link embeddings , we use the speedy , skip-gram neural language model of mikolov et al via their toolkit word2vec .	we use the publicly available 300-dimensional word vectors of mikolov et al , trained on part of the google news dataset .
huang et al use hand-crafted features with lstms to improve performance .	huang et al exploit bilstm to extract features and feed them into crf decoder .
with the fixed order strategy , it performs better if the same strategy is used for evaluation .	in contrast , when the model is trained with the fixed order strategy , it performs better if the same strategy is used for evaluation .
we use 300d glove vectors trained on 840b tokens as the word embedding input to the lstm .	we use glove vectors with 200 dimensions as pre-trained word embeddings , which are tuned during training .
we use the wrapper of the scikit learn python library over the liblinear logistic regression implementation .	we used a logistic regression classifier provided by the liblinear software .
we use the svm implementation available in the li-blinear package .	we use the logistic regression implementation of liblinear wrapped by the scikit-learn library .
grammars can significantly increase the diversity of base models , which plays a central role in parser ensemble , and therefore lead to better and more promising hybrid systems .	moreover , pseudo grammars increase the diversity of base models ; therefore , together with all other models , further improve system combination .
we then used word2vec to train word embeddings with 512 dimensions on each of the prepared corpora .	since our dataset is not so large , we make use of pre-trained word embeddings , which are trained on a much larger corpus with word2vec toolkit .
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 identifying all mentions which refer to the same entity in a document .
in this paper we investigate the use of character-level translation models to support the translation from and to under-resourced languages .	in this paper , we have discussed possibilities to translate via pivot languages on the character level .
we use srilm for training the 5-gram language model with interpolated modified kneser-ney discounting .	we also use a 4-gram language model trained using srilm with kneser-ney smoothing .
the newer method of latent semantic indexing 1 is a variant of the vsm in which documents are represented in a lower dimensional space created from the input training dataset .	latent semantic indexing 1 is a variant of the vsm in which documents are represented in a lower dimensional vector space created from a training dataset .
sentiment analysis is the task of identifying positive and negative opinions , sentiments , emotions and attitudes expressed in text .	sentiment analysis is a collection of methods and algorithms used to infer and measure affection expressed by a writer .
pcfg parsing features are generated on the output of the berkeley parser trained over an english , a german and a spanish treebank .	pcfg parsing features were generated on the output of the berkeley parser , trained over an english and a spanish treebank .
table 1 shows an example item designed for teaching english .	see table 1 for the item produced from the bottom sentence .
all the feature weights and the weight for each probability factor are tuned on the development set with minimum-error-rate training .	the feature weights for the log-linear combination of the features are tuned using minimum error rate training on the devset in terms of bleu .
in § 3 , we describe our approach to paraphrase identification .	in §3 , we describe our approach to paraphrase identification using mt metrics as features .
the penn treebank is perhaps the most influential resource in natural language processing .	the penn treebank is an example of such a resource with worldwide impact on natural language processing .
bilingual data are critical resources for building many applications , such as machine translation and cross language information retrieval .	parallel bilingual corpora are critical resources for statistical machine translation , and cross-lingual information retrieval .
table 1 shows the translation performance by bleu .	the bleu score for all the methods is summarised in table 5 .
the lexicalized reordering models have become the de facto standard in modern phrase-based systems .	among them , lexicalized reordering models have been widely used in practical phrase-based systems .
coreference resolution is the task of determining whether two or more noun phrases refer to the same entity in a text .	coreference resolution is the task of clustering referring expressions in a text so that each resulting cluster represents an entity .
we demonstrate that concept drift is a real , pervasive issue for learning from issue .	we demonstrate that concept drift is an important consideration .
we use hmm alignments along with higher quality alignments from a supervised aligner .	to extract phrases we use hmm alignments along with higher quality alignments from a supervised aligner .
to train our neural algorithm , we apply word embeddings of a look-up from 100-d glove pre-trained on wikipedia and gigaword .	we use 100-dimension glove vectors which are pre-trained on a large twitter corpus and fine-tuned during training .
in recent years , the research community has noticed the great success of neural networks in computer vision , speech recognition and natural language processing tasks .	recently , with the development of neural network , deep learning based models attract much attention in various tasks .
table 4 show the feature templates of our parser , most of which are based on those of zhang and nivre .	our baseline parser uses the feature set described by zhang and nivre .
for any pcfg math-w-7-1-0-40 , there are equivalent ppdts .	a pcfg is proper if math-w-3-1-3-40 for each math-w-3-1-3-55 .
in this paper , we introduced a framework of matrix co-factorization .	in this paper , we address the above challenges with a framework of matrix co-factorization .
word sense disambiguation ( wsd ) is the process of assigning a meaning to a word based on the context in which it occurs .	word sense disambiguation ( wsd ) is the process of determining which sense of a homograph is used in a given context .
we trained a 4-gram language model with kneser-ney smoothing and unigram caching using the sri-lm toolkit .	we estimated 5-gram language models using the sri toolkit with modified kneser-ney smoothing .
which combines the advantages of nmt and smt efficiently .	it is therefore a promising direction to combine the advantages of both nmt and smt .
event coreference resolution is the task of identifying event mentions and clustering them such that each cluster represents a unique real world event .	event coreference resolution is the task of determining which event mentions in a text refer to the same real-world event .
a method based on singular value decomposition provides an efficient and exact solution to this problem .	fortunately , a method based on singular value decomposition provides an efficient and exact solution to this problem .
the input to the network is the embeddings of words , and we use the pre-trained word embeddings by using word2vec on the wikipedia corpus whose size is over 11g .	thus , we pre-train the embeddings on a huge unlabeled data , the chinese wikipedia corpus , with word2vec toolkit .
in addition , we use an english corpus of roughly 227 million words to build a target-side 5-gram language model with srilm in combination with kenlm .	we used kenlm with srilm to train a 5-gram language model based on all available target language training data .
at evaluation time , by proposing a correction only when the confidence of the classifier is high enough , but the article can not be used in training .	cite-p-16-1-11 use the source article at evaluation time and propose a correction only when the score of the classifier is high enough , but the source article is not used in training .
socher et al learned vector space representations for multi-word phrases using recursive autoencoders for the task of sentiment analysis .	socher et al and socher et al present a framework based on recursive neural networks that learns vector space representations for multi-word phrases and sentences .
in both pre-training and fine-tuning , we adopt adagrad and l2 regularizer for optimization .	we use stochastic gradient descent with adagrad , l 2 regularization and minibatch training .
we use opennmt 1 to train the nmt models discussed in this paper .	we use an nmt-small model from the opennmt framework for the neural translation .
as for the boundary detection problem , we use the windowdiff and p k metrics .	we evaluate using the standard penalty metrics p k and windowdiff .
we used a 5-gram language model with modified kneser-ney smoothing implemented using the srilm toolkit .	we implemented this model using the srilm toolkit with the modified kneser-ney discounting and interpolation options .
coreference resolution is a fundamental component of natural language processing ( nlp ) and has been widely applied in other nlp tasks ( cite-p-15-3-9 ) .	coreference resolution is a well known clustering task in natural language processing .
in different relations , ctransr clusters diverse head-tail entity pairs into groups and sets a relation vector for each group .	ctransr is an extension of transr by clustering diverse head-tail entity pairs into groups and learning distinct relation vectors for each group .
the srilm toolkit was used to build this language model .	the model was built using the srilm toolkit with backoff and good-turing smoothing .
to rerank the candidate texts , we used a 5-gram language model trained on the europarl corpus using kenlm .	for building our ap e b2 system , we set a maximum phrase length of 7 for the translation model , and a 5-gram language model was trained using kenlm .
ritter et al learn conversation-specific language models to filter out content words .	ritter et al study twitter dialogues using a clustering approach .
predicates like endocytosis , exocytosis and translocate , though common in biomedical text , are absent from both the framenet and propbank data .	predicates like endocytosis and translocate , though common in biomedical text , are absent from both the framenet and propbank data .
the lstm word embeddings are initialized with 100-dim embeddings from glove and fine-tuned during training .	the word embeddings are initialized using the pre-trained glove , and the embedding size is 300 .
turney used mutual information to choose the best answer to questions about near-synonyms in the test of english as a foreign language and english as a second language .	turney used mutual information to detect the best answer to questions about synonyms from test of english as a foreign language and english as a second language .
we used minimum error rate training mert for tuning the feature weights .	we use our reordering model for n-best re-ranking and optimize bleu using minimum error rate training .
we use the same features as in the first-order model implemented in the mstparser system for syntactic dependency parsing .	we use the arc-based features of turboparser , which descend from several other feature models from the literature on syntactic dependency parsing .
for all experiments , we used a 4-gram language model with modified kneser-ney smoothing which was trained with the srilm toolkit .	we used a 4-gram language model which was trained on the xinhua section of the english gigaword corpus using the srilm 4 toolkit with modified kneser-ney smoothing .
information extraction ( ie ) is a main nlp aspects for analyzing scientific papers , which includes named entity recognition ( ner ) and relation extraction ( re ) .	information extraction ( ie ) is the process of identifying events or actions of interest and their participating entities from a text .
the translation quality is evaluated by case-insensitive bleu-4 metric .	translation quality is evaluated by case-insensitive bleu-4 metric .
coreference resolution is the process of linking together multiple referring expressions of a given entity in the world .	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 .
poon and domingos present an unsupervised semantic parsing approach to partition dependency trees into meaningful fragments .	poon and domingos proposed a model for unsupervised semantic parsing that transforms dependency trees into semantic representations using markov logic .
maltparser is a languageindependent system for data-driven dependency parsing , based on a transition-based parsing model .	maltparser is a data-driven parser-generator , which can induce a dependency parser from a treebank , and which supports several parsing algorithms and learning algorithms .
we used 300-dimensional pre-trained glove word embeddings .	we use theano and pretrained glove word embeddings .
a 5-gram language model with kneser-ney smoothing is trained using s-rilm on the target language .	the language model is a trigram model with modified kneser-ney discounting and interpolation .
evaluation shows that our sentence extraction method performs better than a baseline of taking the sentence with the strongest sentiment .	we showed that our keyphrase-based system performs better than a baseline of extracting the sentence with the highest sentiment score .
we use the official rouge tool 2 to evaluate the performance of the baselines as well as our approach .	we evaluate our models with the standard rouge metric and obtain rouge scores using the pyrouge package .
we have designed a generalized ie system that allows utilizing any tagging strategy .	we have also introduced a new tagging strategy , bia ( begin/after tagging ) .
we also use mini-batch adagrad for optimization and apply dropout .	we apply online training , where model parameters are optimized by using adagrad .
we use a pbsmt model built with the moses smt toolkit .	we use the moses software package 5 to train a pbmt model .
we argue that crime drama exemplified in television programs such as csi : crime scene investigation is an ideal testbed for approximating real-world natural language understanding and the complex inferences associated with it .	specifically , we argue that crime drama exemplified in television programs such as csi : crime scene investigation can be used to approximate real-world natural language understanding and the complex inferences associated with it .
collobert et al adjust the feature embeddings according to the specific task in a deep neural network architecture .	collobert et al use a convolutional neural network over the sequence of word embeddings .
sentiment analysis is the task of identifying the polarity ( positive , negative or neutral ) of review .	the sentiment analysis is a field of study that investigates feelings present in texts .
temporal importance weighting worked very well with textrank .	temporal importance weighting offers consistent improvements over baseline systems .
we build an open-vocabulary language model with kneser-ney smoothing using the srilm toolkit .	in the case of the trigram model , we expand the lattice with the aid of the srilm toolkit .
the resulting constituent parse trees were converted into stanford dependency graphs .	the resulting phrase structures were then converted into dependency structures with the stanford conversion tool .
translating a source language training corpus into target language and creating a corpusbased system in target language 3 .	translating test sentences in target language into source language and inputting them into a source language system 2 .
in philosophy and linguistics , it is accepted that negation conveys positive meaning .	in philosophy and linguistics , it is generally accepted that negation conveys positive meaning .
we propose a data-driven approach for generating short children ’ s stories that does not require extensive manual involvement .	we propose a data-driven approach to story generation that does not require extensive manual involvement .
even worse , mikros and argiri showed that many features besides ngrams are significantly correlated with topic , including sentence and token length , readability measures , and word length distributions .	mikros and argiri have shown that many features besides ngrams are significantly correlated with topic , including sentence and token length , readability measures , and word length distributions .
we see that our estimator compares favorably with the best estimator of vocabulary size .	in this work , we propose a novel nonparametric estimator of vocabulary size .
we employ the trick proposed by blitzer et al to select 魏 pivot features to be reconstructed .	following blitzer et al , we consider pivot features that appear more than 50 times in all the domains .
coreference resolution is the task of determining whether two or more noun phrases refer to the same entity in a text .	coreference resolution is the next step on the way towards discourse understanding .
ratinov and roth and turian et al also explored this approach for name tagging .	turian et al applied this method to both named entity recognition and text chunking .
as word vectors the authors use word2vec embeddings trained with the skip-gram model .	to train the link embeddings , we use the speedy , skip-gram neural language model of mikolov et al via their toolkit word2vec .
for all data sets , we trained a 5-gram language model using the sri language modeling toolkit .	we trained a 5-gram language model on the english side of each training corpus using the sri language modeling toolkit .
transliteration is a key building block for multilingual and cross-lingual nlp since it is useful for user-friendly input methods and applications like machine translation and cross-lingual information retrieval .	transliteration is the conversion of a text from one script to another .
the language model is trained with the sri lm toolkit , on all the available french data without the ted data .	the target language model is trained by the sri language modeling toolkit on the news monolingual corpus .
the language model is trained with the sri lm toolkit , on all the available french data without the ted data .	it has been trained with the srilm toolkit on the target side of all the training data .

we use stanford part-of-speech tagger to automatically detect nouns from text .

we use stanford log-linear partof-speech tagger to produce pos tags for the english side .

as a baseline system for our experiments we use the syntax-based component of the moses toolkit .

our implementation of the segment-based imt protocol is based on the moses toolkit .

we use latent dirichlet allocation , or lda , to obtain a topic distribution over conversations .

an effective strategy to cluster words into topics , is latent dirichlet allocation .

in this paper , we present an unsupervised dynamic bayesian model that allows us to model stylistic style accommodation .

in this paper we presented an unsupervised dynamic bayesian modeling approach to modeling speech style accommodation in face-to-face interactions .

coreference resolution is the process of linking together multiple referring expressions of a given entity in the world .

coreference resolution is the next step on the way towards discourse understanding .

our evaluation on two wordnet-derived taxonomies shows that the learned taxonomies capture a higher number of correct taxonomic relations compared to those produced by traditional distributional similarity approaches .

we evaluate our method on two wordnetderived subtaxonomies and show that our method leads to the development of concept hierarchies that capture a higher number of correct taxonomic relations in comparison to those generated by current distributional similarity approaches .

for training our system classifier , we have used scikit-learn .

we implemented the different aes models using scikit-learn .

we trained a trigram language model on the chinese side , with the srilm toolkit , using the modified kneser-ney smoothing option .

we trained a 4-gram language model on this data with kneser-ney discounting using srilm .

the word embeddings are pre-trained by skip-gram .

all word vectors are trained on the skipgram architecture .

our dataset and parser can be found at http : / / www .

the dataset and parser can be found at http : //www .

parsing is a computationally intensive task due to the combinatorial explosion seen in chart parsing algorithms that explore possible parse trees .

parsing is the process of mapping sentences to their syntactic representations .

through the method , various kinds of collocations induced by key strings are retrieved .

through the method , various range of collocations which are frequently used in a specific domain are retrieved automatically .

mcclosky et al use self-training in combination with a pcfg parser and reranking .

mcclosky et al , 2006 , presents a successful instance of parsing with self-training by using a re-ranker .

our model is a first order linear chain conditional random field .

our model is a structured conditional random field .

rating scales are common in psychology and related fields , such studies hardly exist in nlp , and so .

studies assessing rating scales are very common in psychology and related fields , but are rare in nlp .

we used data from the conll-x shared task on multilingual dependency parsing .

for monolingual treebank data we relied on the conll-x and conll-2007 shared tasks on dependency parsing .

to train the link embeddings , we use the speedy , skip-gram neural language model of mikolov et al via their toolkit word2vec .

we use the publicly available 300-dimensional word vectors of mikolov et al , trained on part of the google news dataset .

huang et al use hand-crafted features with lstms to improve performance .

huang et al exploit bilstm to extract features and feed them into crf decoder .

with the fixed order strategy , it performs better if the same strategy is used for evaluation .

in contrast , when the model is trained with the fixed order strategy , it performs better if the same strategy is used for evaluation .

we use 300d glove vectors trained on 840b tokens as the word embedding input to the lstm .

we use glove vectors with 200 dimensions as pre-trained word embeddings , which are tuned during training .

we use the wrapper of the scikit learn python library over the liblinear logistic regression implementation .

we used a logistic regression classifier provided by the liblinear software .

we use the svm implementation available in the li-blinear package .

we use the logistic regression implementation of liblinear wrapped by the scikit-learn library .

grammars can significantly increase the diversity of base models , which plays a central role in parser ensemble , and therefore lead to better and more promising hybrid systems .

moreover , pseudo grammars increase the diversity of base models ; therefore , together with all other models , further improve system combination .

we then used word2vec to train word embeddings with 512 dimensions on each of the prepared corpora .

since our dataset is not so large , we make use of pre-trained word embeddings , which are trained on a much larger corpus with word2vec toolkit .

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 identifying all mentions which refer to the same entity in a document .

in this paper we investigate the use of character-level translation models to support the translation from and to under-resourced languages .

in this paper , we have discussed possibilities to translate via pivot languages on the character level .

we use srilm for training the 5-gram language model with interpolated modified kneser-ney discounting .

we also use a 4-gram language model trained using srilm with kneser-ney smoothing .

the newer method of latent semantic indexing 1 is a variant of the vsm in which documents are represented in a lower dimensional space created from the input training dataset .

latent semantic indexing 1 is a variant of the vsm in which documents are represented in a lower dimensional vector space created from a training dataset .

sentiment analysis is the task of identifying positive and negative opinions , sentiments , emotions and attitudes expressed in text .

sentiment analysis is a collection of methods and algorithms used to infer and measure affection expressed by a writer .

pcfg parsing features are generated on the output of the berkeley parser trained over an english , a german and a spanish treebank .

pcfg parsing features were generated on the output of the berkeley parser , trained over an english and a spanish treebank .

table 1 shows an example item designed for teaching english .

see table 1 for the item produced from the bottom sentence .

all the feature weights and the weight for each probability factor are tuned on the development set with minimum-error-rate training .

the feature weights for the log-linear combination of the features are tuned using minimum error rate training on the devset in terms of bleu .

in § 3 , we describe our approach to paraphrase identification .

in §3 , we describe our approach to paraphrase identification using mt metrics as features .

the penn treebank is perhaps the most influential resource in natural language processing .

the penn treebank is an example of such a resource with worldwide impact on natural language processing .

bilingual data are critical resources for building many applications , such as machine translation and cross language information retrieval .

parallel bilingual corpora are critical resources for statistical machine translation , and cross-lingual information retrieval .

table 1 shows the translation performance by bleu .

the bleu score for all the methods is summarised in table 5 .

the lexicalized reordering models have become the de facto standard in modern phrase-based systems .

among them , lexicalized reordering models have been widely used in practical phrase-based systems .

coreference resolution is the task of determining whether two or more noun phrases refer to the same entity in a text .

coreference resolution is the task of clustering referring expressions in a text so that each resulting cluster represents an entity .

we demonstrate that concept drift is a real , pervasive issue for learning from issue .

we demonstrate that concept drift is an important consideration .

we use hmm alignments along with higher quality alignments from a supervised aligner .

to extract phrases we use hmm alignments along with higher quality alignments from a supervised aligner .

to train our neural algorithm , we apply word embeddings of a look-up from 100-d glove pre-trained on wikipedia and gigaword .

we use 100-dimension glove vectors which are pre-trained on a large twitter corpus and fine-tuned during training .

in recent years , the research community has noticed the great success of neural networks in computer vision , speech recognition and natural language processing tasks .

recently , with the development of neural network , deep learning based models attract much attention in various tasks .

table 4 show the feature templates of our parser , most of which are based on those of zhang and nivre .

our baseline parser uses the feature set described by zhang and nivre .

for any pcfg math-w-7-1-0-40 , there are equivalent ppdts .

a pcfg is proper if math-w-3-1-3-40 for each math-w-3-1-3-55 .

in this paper , we introduced a framework of matrix co-factorization .

in this paper , we address the above challenges with a framework of matrix co-factorization .

word sense disambiguation ( wsd ) is the process of assigning a meaning to a word based on the context in which it occurs .

word sense disambiguation ( wsd ) is the process of determining which sense of a homograph is used in a given context .

we trained a 4-gram language model with kneser-ney smoothing and unigram caching using the sri-lm toolkit .

we estimated 5-gram language models using the sri toolkit with modified kneser-ney smoothing .

which combines the advantages of nmt and smt efficiently .

it is therefore a promising direction to combine the advantages of both nmt and smt .

event coreference resolution is the task of identifying event mentions and clustering them such that each cluster represents a unique real world event .

event coreference resolution is the task of determining which event mentions in a text refer to the same real-world event .

a method based on singular value decomposition provides an efficient and exact solution to this problem .

fortunately , a method based on singular value decomposition provides an efficient and exact solution to this problem .

the input to the network is the embeddings of words , and we use the pre-trained word embeddings by using word2vec on the wikipedia corpus whose size is over 11g .

thus , we pre-train the embeddings on a huge unlabeled data , the chinese wikipedia corpus , with word2vec toolkit .

in addition , we use an english corpus of roughly 227 million words to build a target-side 5-gram language model with srilm in combination with kenlm .

we used kenlm with srilm to train a 5-gram language model based on all available target language training data .

at evaluation time , by proposing a correction only when the confidence of the classifier is high enough , but the article can not be used in training .

cite-p-16-1-11 use the source article at evaluation time and propose a correction only when the score of the classifier is high enough , but the source article is not used in training .

socher et al learned vector space representations for multi-word phrases using recursive autoencoders for the task of sentiment analysis .

socher et al and socher et al present a framework based on recursive neural networks that learns vector space representations for multi-word phrases and sentences .

in both pre-training and fine-tuning , we adopt adagrad and l2 regularizer for optimization .

we use stochastic gradient descent with adagrad , l 2 regularization and minibatch training .

we use opennmt 1 to train the nmt models discussed in this paper .

we use an nmt-small model from the opennmt framework for the neural translation .

as for the boundary detection problem , we use the windowdiff and p k metrics .

we evaluate using the standard penalty metrics p k and windowdiff .

we used a 5-gram language model with modified kneser-ney smoothing implemented using the srilm toolkit .

we implemented this model using the srilm toolkit with the modified kneser-ney discounting and interpolation options .

coreference resolution is a fundamental component of natural language processing ( nlp ) and has been widely applied in other nlp tasks ( cite-p-15-3-9 ) .

coreference resolution is a well known clustering task in natural language processing .

in different relations , ctransr clusters diverse head-tail entity pairs into groups and sets a relation vector for each group .

ctransr is an extension of transr by clustering diverse head-tail entity pairs into groups and learning distinct relation vectors for each group .

the srilm toolkit was used to build this language model .

the model was built using the srilm toolkit with backoff and good-turing smoothing .

to rerank the candidate texts , we used a 5-gram language model trained on the europarl corpus using kenlm .

for building our ap e b2 system , we set a maximum phrase length of 7 for the translation model , and a 5-gram language model was trained using kenlm .

ritter et al learn conversation-specific language models to filter out content words .

ritter et al study twitter dialogues using a clustering approach .

predicates like endocytosis , exocytosis and translocate , though common in biomedical text , are absent from both the framenet and propbank data .

predicates like endocytosis and translocate , though common in biomedical text , are absent from both the framenet and propbank data .

the lstm word embeddings are initialized with 100-dim embeddings from glove and fine-tuned during training .

the word embeddings are initialized using the pre-trained glove , and the embedding size is 300 .

turney used mutual information to choose the best answer to questions about near-synonyms in the test of english as a foreign language and english as a second language .

turney used mutual information to detect the best answer to questions about synonyms from test of english as a foreign language and english as a second language .

we used minimum error rate training mert for tuning the feature weights .

we use our reordering model for n-best re-ranking and optimize bleu using minimum error rate training .

we use the same features as in the first-order model implemented in the mstparser system for syntactic dependency parsing .

we use the arc-based features of turboparser , which descend from several other feature models from the literature on syntactic dependency parsing .

for all experiments , we used a 4-gram language model with modified kneser-ney smoothing which was trained with the srilm toolkit .

we used a 4-gram language model which was trained on the xinhua section of the english gigaword corpus using the srilm 4 toolkit with modified kneser-ney smoothing .

information extraction ( ie ) is a main nlp aspects for analyzing scientific papers , which includes named entity recognition ( ner ) and relation extraction ( re ) .

information extraction ( ie ) is the process of identifying events or actions of interest and their participating entities from a text .

the translation quality is evaluated by case-insensitive bleu-4 metric .

translation quality is evaluated by case-insensitive bleu-4 metric .

coreference resolution is the process of linking together multiple referring expressions of a given entity in the world .

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 .

poon and domingos present an unsupervised semantic parsing approach to partition dependency trees into meaningful fragments .

poon and domingos proposed a model for unsupervised semantic parsing that transforms dependency trees into semantic representations using markov logic .

maltparser is a languageindependent system for data-driven dependency parsing , based on a transition-based parsing model .

maltparser is a data-driven parser-generator , which can induce a dependency parser from a treebank , and which supports several parsing algorithms and learning algorithms .

we used 300-dimensional pre-trained glove word embeddings .

we use theano and pretrained glove word embeddings .

a 5-gram language model with kneser-ney smoothing is trained using s-rilm on the target language .

the language model is a trigram model with modified kneser-ney discounting and interpolation .

evaluation shows that our sentence extraction method performs better than a baseline of taking the sentence with the strongest sentiment .

we showed that our keyphrase-based system performs better than a baseline of extracting the sentence with the highest sentiment score .

we use the official rouge tool 2 to evaluate the performance of the baselines as well as our approach .

we evaluate our models with the standard rouge metric and obtain rouge scores using the pyrouge package .

we have designed a generalized ie system that allows utilizing any tagging strategy .

we have also introduced a new tagging strategy , bia ( begin/after tagging ) .

we also use mini-batch adagrad for optimization and apply dropout .

we apply online training , where model parameters are optimized by using adagrad .

we use a pbsmt model built with the moses smt toolkit .

we use the moses software package 5 to train a pbmt model .

we argue that crime drama exemplified in television programs such as csi : crime scene investigation is an ideal testbed for approximating real-world natural language understanding and the complex inferences associated with it .

specifically , we argue that crime drama exemplified in television programs such as csi : crime scene investigation can be used to approximate real-world natural language understanding and the complex inferences associated with it .

collobert et al adjust the feature embeddings according to the specific task in a deep neural network architecture .

collobert et al use a convolutional neural network over the sequence of word embeddings .

sentiment analysis is the task of identifying the polarity ( positive , negative or neutral ) of review .

the sentiment analysis is a field of study that investigates feelings present in texts .

temporal importance weighting worked very well with textrank .

temporal importance weighting offers consistent improvements over baseline systems .

we build an open-vocabulary language model with kneser-ney smoothing using the srilm toolkit .

in the case of the trigram model , we expand the lattice with the aid of the srilm toolkit .

the resulting constituent parse trees were converted into stanford dependency graphs .

the resulting phrase structures were then converted into dependency structures with the stanford conversion tool .

translating a source language training corpus into target language and creating a corpusbased system in target language 3 .

translating test sentences in target language into source language and inputting them into a source language system 2 .

in philosophy and linguistics , it is accepted that negation conveys positive meaning .

in philosophy and linguistics , it is generally accepted that negation conveys positive meaning .

we propose a data-driven approach for generating short children ’ s stories that does not require extensive manual involvement .

we propose a data-driven approach to story generation that does not require extensive manual involvement .

even worse , mikros and argiri showed that many features besides ngrams are significantly correlated with topic , including sentence and token length , readability measures , and word length distributions .

mikros and argiri have shown that many features besides ngrams are significantly correlated with topic , including sentence and token length , readability measures , and word length distributions .

we see that our estimator compares favorably with the best estimator of vocabulary size .

in this work , we propose a novel nonparametric estimator of vocabulary size .

we employ the trick proposed by blitzer et al to select 魏 pivot features to be reconstructed .

following blitzer et al , we consider pivot features that appear more than 50 times in all the domains .

coreference resolution is the task of determining whether two or more noun phrases refer to the same entity in a text .

coreference resolution is the next step on the way towards discourse understanding .

ratinov and roth and turian et al also explored this approach for name tagging .

turian et al applied this method to both named entity recognition and text chunking .

as word vectors the authors use word2vec embeddings trained with the skip-gram model .

to train the link embeddings , we use the speedy , skip-gram neural language model of mikolov et al via their toolkit word2vec .

for all data sets , we trained a 5-gram language model using the sri language modeling toolkit .

we trained a 5-gram language model on the english side of each training corpus using the sri language modeling toolkit .

transliteration is a key building block for multilingual and cross-lingual nlp since it is useful for user-friendly input methods and applications like machine translation and cross-lingual information retrieval .

transliteration is the conversion of a text from one script to another .

the language model is trained with the sri lm toolkit , on all the available french data without the ted data .

the target language model is trained by the sri language modeling toolkit on the news monolingual corpus .

the language model is trained with the sri lm toolkit , on all the available french data without the ted data .

it has been trained with the srilm toolkit on the target side of all the training data .