text
stringlengths
82
736
label
int64
0
1
in this experiment we have customized freely available maltparser which follows a data-driven approach---rosti et al described an incremental ter alignment to mitigate these problems
0
ma et al proposed interactive attention network which interactively learns attentions in the contexts and targets , and generates the representations for targets and contexts separately---ma et al further proposed bidirectional attention mechanism , which also learns the attention weights on aspect words towards the averaged vector of context words
1
the word vectors of vocabulary words are trained from a large corpus using the glove toolkit---we use the pre-trained glove 50-dimensional word embeddings to represent words found in the glove dataset
1
some of the recent works that have employed pre-trained language models include ulmfit , elmo , glomo , bert and openai transformer---lee et al presented an end-to-end coreference resolution model which reasons over all the anteceding spans
0
the approach learns from a small , annotated corpus and the task includes resolving not just pronouns but general noun phrases---and the task includes resolving not just a certain type of noun phrase ( e . g . , pronouns )
1
we propose using a principled way of incorporating both rater-comment and rater-author interactions simultaneously---to this end , we propose a factor model that incorporates rater-comment and rater-author interactions simultaneously
1
relation extraction is a challenging task in natural language processing---relation extraction ( re ) is the task of recognizing relationships between entities mentioned in text
1
we use binary cross-entropy as the objective function and the adam optimization algorithm with the parameters suggested by kingma and ba for training the network---we train the classifier with log-loss and adam optimization algorithm , including dropout and early stopping for regularization
1
semantic role labeling ( srl ) is the task of identifying the semantic arguments of a predicate and labeling them with their semantic roles---semantic role labeling ( srl ) is the task of labeling predicate-argument structure in sentences with shallow semantic information
1
we primarily compared our model with conditional random fields---our system is based on the conditional random field
1
word sense disambiguation ( wsd ) is a particular problem of computational linguistics which consists in determining the correct sense for a given ambiguous word---word sense disambiguation ( wsd ) is a problem long recognised in computational linguistics ( yngve 1955 ) and there has been a recent resurgence of interest , including a special issue of this journal devoted to the topic ( cite-p-27-8-11 )
1
the robust processing capabilities of the parser are demonstrated in its use in improving the accuracy of a speech recognizer---the network is trained with backpropagation and the gradientbased optimization is performed using the adagrad update rule
0
empirically , s-lstm can give effective sentence encoding after 3 ¨c 6 recurrent steps---in this article , we have considered the task of automatically generating questions from topics
0
therefore , we generate 50-best hypothesis from the ensemble system and then tune the model weights with batch-mira on the development set to maximize the bleu score---we evaluate the mt system with the bleu metric , papineni et al , 2002 decisions
0
topic models such as latent dirichlet allocation are hierarchical probabilistic models of document collections---statistical topic models such as latent dirichlet allocation provide a powerful framework for representing and summarizing the contents of large document collections
1
a discourse consists of a sequence of utterances uttl , ... , uttn---discourse is a structurally organized set of coherent text segments
1
we evaluated the translation quality of the system using the bleu metric---we evaluate the performance of different translation models using both bleu and ter metrics
1
for simplicity , we use the well-known conditional random fields for sequential labeling---we solve this sequence tagging problem using the mallet implementation of conditional random fields
1
twitter is a well-known social network service that allows users to post short 140 character status update which is called “ tweet ”---twitter is a social platform which contains rich textual content
1
discourse parsing is a challenging task and is crucial for discourse analysis---discourse parsing is a difficult , multifaceted problem involving the understanding and modeling of various semantic and pragmatic phenomena as well as understanding the structural properties that a discourse graph can have
1
for the fluency and grammaticality features , we train 4-gram lms using the development dataset with the sri toolkit---we build an open-vocabulary language model with kneser-ney smoothing using the srilm toolkit
1
we used the 200-dimensional word vectors for twitter produced by glove---we use glove vectors with 100 dimensions trained on wikipedia and gigaword as word embeddings
1
the values of the word embeddings matrix e are learned using the neural network model introduced by---we further used a 5-gram language model trained using the srilm toolkit with modified kneser-ney smoothing
0
we used the treetagger tool to extract part-of-speech from each given text , then tokenize and lemmatize it---sarcasm is a sophisticated form of communication in which speakers convey their message in an indirect way
0
to solve this dynamic state tracking problem , we propose a sequential labeling approach using linear-chain conditional random fields---we solve this sequence tagging problem using the mallet implementation of conditional random fields
1
the most common word embeddings used in deep learning are word2vec , glove , and fasttext---semi-supervised learning is a machine learning approach that utilizes large amounts of unlabeled data , combined with a smaller amount of labeled data , to learn a target function
0
here we use stanford corenlp toolkit to deal with the co-reference problem---we used the dependency parser from the stanford corenlp
1
we then lowercase all data and use all sentences from the modern dutch part of the corpus to train an n-gram language model with the srilm toolkit---we train a trigram language model with modified kneser-ney smoothing from the training dataset using the srilm toolkit , and use the same language model for all three systems
1
we measure the translation quality with automatic metrics including bleu and ter---neural networks have recently gained much attention as a way of inducing word vectors
0
a 5-gram language model with kneser-ney smoothing is trained using s-rilm on the target language---a kn-smoothed 5-gram language model is trained on the target side of the parallel data with srilm
1
the ukwac web-corpus is used as a native corpus for training the suggestion model---however , in , as the difficulty shown in the experiments , the whole sentiment of a document is not necessarily the sum of its parts
0
we proposed a new method for translation acquisition which uses a set of synonyms to acquire translations---we propose a new method for translation acquisition which uses a set of synonyms to acquire translations
1
gram language models are trained over the target-side of the training data , using srilm with modified kneser-ney discounting---our model is based on the standard lstm encoder-decoder model with an attention mechanism
0
for language model , we used sri language modeling toolkit to train a 4-gram model with modified kneser-ney smoothing---in order to limit the size of the vocabulary of the unmt model , we segmented tokens in the training data into sub-word units via byte pair encoding
0
we perform an analysis of humans ’ perceptions of formality in four different genres---we provide an analysis of humans ’ subjective perceptions of formality in four different genres
1
recently , mikolov et al introduced an efficient way for inferring word embeddings that are effective in capturing syntactic and semantic relationships in natural language---the pol-yglot project mikolov et al developed an alternative solution for computing word embeddings , which significantly reduces the computational costs
1
in this paper we present an unsupervised approach to relational information extraction---in this paper we presented a new model for unsupervised relation extraction which operates over tuples
1
in this paper , we make use of curated monolingual linguistic resources in the source side to improve nmt in bilingually scarce scenarios---we used datasets distributed for the 2006 and 2007 conll shared tasks
0
we use the stanford pos tagger to obtain the lemmatized corpora for the sre task---for all pos tagging tasks we use the stanford log-linear part-ofspeech tagger
1
we use the well-known word embedding model that is a robust framework to incorporate word representation features---we use a cws-oriented model modified from the skip-gram model to derive word embeddings
1
coreference resolution is the task of grouping mentions to entities---桅 n is similar in form to the weighted finite-state transducer representation of a backoff n-gram language model
0
all corpora were preprocessed using the standard moses scripts to perform normalization , tokenization , and truecasing---the pipeline consisted in normalizing punctuation , tokenization and truecasing using the standard moses scripts
1
sentiment analysis is a much-researched area that deals with identification of positive , negative and neutral opinions in text---sentiment analysis is a research area where does a computational analysis of people ’ s feelings or beliefs expressed in texts such as emotions , opinions , attitudes , appraisals , etc . ( cite-p-12-1-3 )
1
a linear context-free rewriting system is a linear , non-erasing multiple context-free grammar---a stochastic multiple context-free grammar is a probabilistic extension of mcfg or linear context-free rewriting system
1
qiu et al propose a double propagation method to extract opinion word and opinion target simultaneously---qiu et al proposed double propagation to collectively extract aspect terms and opinion words based on information propagation over a dependency graph
1
the score combination weights are trained by a minimum error rate training procedure similar to---the feature weights 位 i are trained in concert with the lm weight via minimum error rate training
1
the demo is available at http : //twine-mind.cloudapp.net/streaming 1,2---you can try the demo at http : / / twine-mind . cloudapp . net / streaming-demo
1
part-of-speech ( pos ) tagging is a fundamental natural-language-processing problem , and pos tags are used as input to many important applications---part-of-speech ( pos ) tagging is a fundamental nlp task , used by a wide variety of applications
1
gram language model with modified kneser-ney smoothing is trained with the srilm toolkit on the epps , ted , newscommentary , and the gigaword corpora---the language models are 4-grams with modified kneser-ney smoothing which have been trained with the srilm toolkit
1
relation extraction ( re ) is the task of extracting semantic relationships between entities in text---relation extraction is a crucial task in the field of natural language processing ( nlp )
1
coreference resolution is a challenging task , that involves identification and clustering of noun phrases mentions that refer to the same real-world entity---coreference resolution is the task of grouping all the mentions of entities 1 in a document into equivalence classes so that all the mentions in a given class refer to the same discourse entity
1
our framework was built with the cleartk toolkit with its wrapper for svmlight---system tuning was carried out using both k-best mira and minimum error rate training on the held-out development set
0
topic modeling is an unsupervised learning algorithm that can automatically discover themes of a document collection---topic modeling is an unsupervised technique that can automatically identify themes from a given set of documents
1
we demonstrate superagent as an add-on extension to mainstream web browsers and show its usefulness to user ’ s online shopping experience---we demonstrate superagent as an add-on extension to mainstream web browsers such as microsoft edge and google chrome
1
alshawi et al , 2000 ) represents each production in parallel dependency trees as a finite-state transducer---supervised ner approaches can often achieve high accuracy when a large annotated training set similar to the test data is available
0
knowledge graphs like wordnet , freebase , and dbpedia have become extremely useful resources for many nlp-related applications---knowledge graphs such as freebase , yago and wordnet are among the most widely used resources in nlp applications
1
the language model was constructed using the srilm toolkit with interpolated kneser-ney discounting---for the language model , we used srilm with modified kneser-ney smoothing
1
curran and moens have demonstrated that dramatically increasing the quantity of text used to extract contexts significantly improves synonym quality---schwenk proposed a feed-forward network that computes phrase scores offline , and the scores were added to the phrase table of a phrasebased system
0
one such urll distance measure is given in---one distance measure for urll trees is introduced in
1
the penn discourse treebank is a new resource with annotations of discourse connectives and their senses in the wall street journal portion of the penn treebank---the penn discourse treebank provides annotations for the arguments and relation senses of one hundred pre-selected discourse connectives over the news portion of the penn treebank corpus
1
we here address a particular flexible cg , the lambek calculus---we are concerned with the implicational fragment of the associative lambek calculus
1
it has been empirically shown that word embeddings could capture semantic and syntactic similarities between words---distributed representations for words and sentences have been shown to significantly boost the performance of a nlp system
1
these rules are able to sequentially derive the various alternated forms from a single base form , which is stated in the lexical entry---rules derive the more complex forms from a basic one , which is the only one that needs to be stated in the lexical entry
1
our neural model of ape is based on the work described in cohn et al which implements structural alignment biases into an attention based bidirectional recurrent neural network mt model---inspired by bahdanau et al , our deep neural network model uses a bidirectional recurrent neural network with gated recurrent units
1
our smt system is a phrase-based system based on the moses smt toolkit---the promt smt system is based on the moses open-source toolkit
1
in their interactive alignment model , pickering and garrod suggest that dialogue between humans is greatly aided by aligning representations on several linguistic and conceptual levels---in contrast , pickering and garrod have argued that stimulusresponse priming is the key mechanism underlying alignment of representations in conversation
1
we used bleu for automatic evaluation of our ebmt systems---our focus was on minimizing the bleu score of the development set
1
word embeddings have proven to be effective models of semantic representation of words in various nlp tasks---high quality word embeddings have been proven helpful in many nlp tasks
1
word representations to learn word embeddings from our unlabeled corpus , we use the gensim im-plementation of the word2vec algorithm---we use the skipgram model with negative sampling implemented in the open-source word2vec toolkit to learn word representations
1
semantic parsing is the task of translating natural language utterances into a machine-interpretable meaning representation---semantic parsing is the task of converting natural language utterances into formal representations of their meaning
1
for assessing significance , we apply the approximate randomization test---we propose a third approach , in which the task of invoking responses from the system is treated as one of retrieval from the set of all possible responses
0
we optimise the feature weights of the model with minimum error rate training against the bleu evaluation metric---we used srilm -sri language modeling toolkit to train several character models
0
to minimize noise , we also consider semantically similar tweets posted by other users---in a single tweet , we also model the similar tweets posted by all other users with reinforced inter-user representation
1
we first extend the matrix factorization framework to explicitly model the corresponding relationships between feature classes and examples classes---in matrix factorization , we propose a unified scheme to evaluate the value of feature and example labels
1
relation extraction is the key component for building relation knowledge graphs , and it is of crucial significance to natural language processing applications such as structured search , sentiment analysis , question answering , and summarization---relation extraction ( re ) is a task of identifying typed relations between known entity mentions in a sentence
1
for previous work on constituent parsing , sarkar and han used an early version of the korean penn treebank to train lexicalized tree adjoining grammars---hermjakob implemented a shift-reduce parser for korean trained on very limited data , and sarkar and han used an earlier version of the treebank to train a lexicalized tree adjoining grammar
1
a system pbmt is built using the phrase-based model in moses---our experimental results demonstrate both that our proposed approach is useful in predicting missing preferences of users
0
in this paper , we propose a novel method to model short texts based on semantic clustering and convolutional neural network---in this paper , we design a method to exploit more contextual information for short text classification
1
a 4-gram language model was trained on the monolingual data by the srilm toolkit---sentence compression is the task of producing a summary at the sentence level
0
conditional random fields are discriminative structured classification models for sequential tagging and segmentation---conditional random fields are undirected graphical models that are conditionally trained
1
by leveraging the knowledge extracted from the wikipedia relation repository , our approach significantly improves the performance over the state-of-the-art approaches on ace data---background-knowledge-based topics generated from the wikipedia relation repository can significantly improve the performance over the state-of-the-art relation detection approaches
1
the feature weights are tuned to optimize bleu using the minimum error rate training algorithm---textual entailment is the task of automatically determining whether a natural language hypothesis can be inferred from a given piece of natural language text
0
we apply standard tuning with mert on the bleu score---we report the mt performance using the original bleu metric
1
we used the google news pretrained word2vec word embeddings for our model---for each word belonging to any of our activities , we use wordnet to find its synonyms
0
phrase pairs are built by combining minimal translation units and ordering information---phrase frequencies are obtained by counting all possible occurrences
1
the ukwac web-corpus is used as a native corpus for training the suggestion model---the corpus used for learning feature-norm-like concept descriptions is ukwac
1
for both languages , we used the srilm toolkit to train a 5-gram language model using all monolingual data provided---we used 5-gram models , estimated using the sri language modeling toolkit with modified kneser-ney smoothing
1
the system is evaluated with bleu and then scored for precision , recall , and f1 measure against the dev set reference---for a fair comparison to our model , we used word2vec , that pretrain word embeddings at a token level
0
we use the same datasets as in hoffmann et al and riedel et al , which include 3-years of new york times articles aligned with freebase---we use belief propagation for inference in our crfs
0
berland and charniak proposed a system for part-of relation extraction , based on the approach---berland and charniak proposed a similar method for part-whole relations
1
since the computation of full softmax is time consuming , the techniques of hierarchical softmax and negative sampling are proposed for approximation---to address this issue , several efficient methods have been proposed such as hierarchical softmax tree and negative sampling
1
levinger et al used morpho-lexical probabilities learned from an untagged corpus for morphological disambiguation of hebrew texts---levinger et al developed a method for disambiguation of the results provided by a morphological analyzer for hebrew
1
all features were log-linearly combined and their weights were optimized by performing minimum error rate training---the weights of the different feature functions were tuned by means of minimum error-rate training executed on the europarl development corpus
1
in this paper , the result is improved to o ( n4 log n ) as a new lowest upper bound---so that the expected result should be a linear time bound on o ( n2 )
1
we use the wsj corpus , a pos annotated corpus , for this purpose---we use the penn treebank as the linguistic data source
1
we use kaldi , an open-source speech recognition framework and acoustic models based on the ted-lium corpus and the tedlium 4-gram language model from cantab research---relation extraction ( re ) is the task of determining semantic relations between entities mentioned in text
0
in this treebank , we followed the format of the conll tab-separated format for dependency parsing---pichotta and mooney experimented with lstm for script learning , using an existing sequence of events to predict the probability of a next event , which outperformed strong discrete baselines
0
we trained a trigram language model on the chinese side , with the srilm toolkit , using the modified kneser-ney smoothing option---we use srilm to train a 5-gram language model on the target side of our training corpus with modified kneser-ney discounting
1
we apply this method to english pos tagging and japanese morphological analysis---we apply this method to english part-of-speech tagging and japanese morphological analysis
1
we have shown that active learning strategies can reduce the effort involved in eliciting human alignment data---that has looked at reducing human effort by selective elicitation of partial word alignment using active learning techniques
1