text
stringlengths
82
736
label
int64
0
1
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---we pretrain 200-dimensional word embeddings using word2vec on the english wikipedia corpus , and randomly initialize other hyperparameters
1
a 4-grams language model is trained by the srilm toolkit---the srilm toolkit is used to train 5-gram language model
1
our approach formalizes semantic role induction as a graph partitioning problem---relation extraction ( re ) has been defined as the task of identifying a given set of semantic binary relations in text
0
the weights of the different feature functions were optimised by means of minimum error rate training on the 2013 wmt test set---feature weights were set with minimum error rate training on a development set using bleu as the objective function
1
this paper proposes a method for intra-sentential subject zero anaphora resolution in japanese---the text is parsed using the rasp parser , and subcategorizations are extracted using the system of briscoe and carroll
0
we used a regularized maximum entropy model---for entity tagging we used a maximum entropy model
1
kennedy and hirst proposed a more reliable procedure that leverages two existing aligned monolingual word similarity datasets for the construction of a new cross-lingual dataset---kennedy and hirst proposed a method which exploits two aligned monolingual word similarity datasets for the construction of a french-english cross-lingual dataset
1
existing work has used the masking of random words to build language models as well as contextualized word embeddings---previous research has shown the usefulness of using pretrained word vectors to improve the performance of various models
1
we used a 5-gram language model with modified kneser-ney smoothing , built with the srilm toolkit---word re-embedding based on manifold learning can help the original space
0
the x-lingual method uses unlabeled parallel sentences to learn crosslingual word clusters and used them as augmenting features to train a delexicalized mstparser---the x-lingual method uses unlabeled parallel sentences to induce cross-lingual word clusters as augmenting features for delexicalized dependency parser
1
thus , the main contribution of our work is to propose a relative entropy pruning model for translation models used in phrase-based machine translation---in this work , we propose a relative entropy model for translation models , that measures how likely a phrase pair encodes a translation event that is derivable
1
the language model was a kneser-ney interpolated trigram model generated using the srilm toolkit---the language model is a trigram model with modified kneser-ney discounting and interpolation
1
in a semantic role labeling task , the syntax and semantics are correlated with each other , that is , the global structure of the sentence is useful for identifying ambiguous semantic roles---in a language understanding task , the head word dependencies or parse tree path are successfully applied to learn and predict semantic roles , especially those with ambiguous labels
1
we used a 5-gram language model with modified kneser-ney smoothing implemented using the srilm toolkit---we used the srilm toolkit to simulate the behavior of flexgram models by using count files as input
1
in this paper we investigate the applicability of co-training to train classifiers that predict emotions in spoken dialogues---we investigate the automatic labeling of spoken dialogue data , in order to train a classifier that predicts students ¡¯ emotional states
1
we add gaussian noise at the embedding layer and use dropout to ignore the signal from a set of randomly selected neurons in the network---we apply dropout on the lstm layer to prevent network parameters from overfitting and control the co-adaptation of features
1
we use an information extraction tool for named entity recognition based on conditional random fields---we use a conditional random field sequence model , which allows for globally optimal training and decoding
1
we use the multi-class logistic regression classifier from the liblinear package 2 for the prediction of edit scripts---we applied liblinear via its scikitlearn python interface to train the logistic regression model with l2 regularization
1
the lr and svm classifiers were implemented with scikit-learn---the standard classifiers are implemented with scikit-learn
1
text reuse is the process of creating new document ( s ) using text from existing document ( s )---text reuse is the transformation of a source text into a target text in order to serve a different purpose
1
the language model is a trigram-based backoff language model with kneser-ney smoothing , computed using srilm and trained on the same training data as the translation model---the language model was a 5-gram language model estimated on the target side of the parallel corpora by using the modified kneser-ney smoothing implemented in the srilm toolkit
1
for this purpose , we use the maximum entropy modeling with inequality constraints---to control overfitting in the maxent models , we used box-type inequality constraints
1
to the best of our knowledge , previous research to apply continuous space methods to the translation model , were limited to tuple-based translation models le et al , 2012 )---previous works on continuous space translation models in an bilingual tuple system only used rescoring le et al , 2012 )
1
in particular , neural language models have demonstrated impressive performance at the task of language modeling---for language model , we used sri language modeling toolkit to train a 4-gram model with modified kneser-ney smoothing
0
madamira is a tool , originally designed for morphological analysis and disambiguation of msa and dialectal arabic texts---madamira is a system developed for morphological analysis and disambiguation of arabic text
1
example retrieval systems such as rakhilina et al and kilgarriff et al particularly check for the appropriate use of words in the context in which they are written---thanks to the constraints on dependency trees , it is possible to reduce complexity to ofor lexicalized parsing using the spanbased representation proposed by eisner
0
chambers and jurafsky proposed a method to learn narrative chains of events related to a protagonist in a single document---chambers and jurafsky introduced the concept of narrative event chains as a representation of structured event relation knowledge
1
human one-to-one tutoring often yields significantly higher learning gains than classroom instruction---one-on-one tutoring has been shown to be a very effective form of instruction
1
relation extraction is the task of finding relations between entities in text , which is useful for several tasks such as information extraction , summarization , and question answering ( cite-p-14-3-7 )---relation extraction is the task of finding semantic relations between entities from text
1
we created a data collection for research into why-questions and for development of a method for why-qa---we created a data collection for research , development and evaluation of a method for automatically answering why-questions ( why-qa )
1
in japanese morphological analysis , the dictionary-based approach has been widely used to generate the word lattice , kurohashi et al ,---in japanese morphological analysis , the dictionary-based approach has been widely used to generate word lattices
1
we employed the glove as the word embedding for the esim---for the neural models , we use 100-dimensional glove embeddings , pre-trained on wikipedia and gigaword
0
for our logistic regression classifier we use the implementation included in the scikit-learn toolkit 2---we use logistic regression with l2 regularization , implemented using the scikit-learn toolkit
1
for all submissions , we used the phrase-based variant of the moses decoder---we used a phrase-based smt model as implemented in the moses toolkit
1
we exploited glove vectors instead of one-hot vectors in order to facilitate generalization---for this reason , we used glove vectors to extract the vector representation of words
1
however , as we demonstrate in sec . 5 , human judgment can result in inconsistent scoring---in sec . 5 , human judgment can result in inconsistent scoring
1
socher et al present a model for compositionality based on recursive neural networks---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
0
we train and evaluate a l2-regularized logistic regression classifier with the liblin-ear solver as implemented in scikit-learn---traditional topic models such as lda and plsa are unsupervised methods for extracting latent topics in text documents
0
we used the target side of the parallel corpus and the srilm toolkit to train a 5-gram language model---for all data sets , we trained a 5-gram language model using the sri language modeling toolkit
1
the translation systems were evaluated by bleu score---the translations are evaluated in terms of bleu score
1
relation extraction is the task of finding semantic relations between two entities from text---relation extraction is the task of automatically detecting occurrences of expressed relations between entities in a text and structuring the detected information in a tabularized form
1
we use the rouge toolkit for evaluation of the generated summaries in comparison to the gold summaries---we have used rouge-1 , which gives good results with standard summaries
1
much current work in discourse parsing focuses on the labelling of discourse relations , using data from the penn discourse treebank---from these experiments , we find that rich and broad information improves the disambiguation performance considerably
0
significance tests are conducted using bootstrap sampling---luong et al preprocess the data and replace each unknown word in the target sentence by a placeholder token also containing a positional pointer to the corresponding word in the source sentence
0
to implement svm algorithm , we have used the publicly available python based scikit-learn package---sentiment analysis is the process of identifying and extracting subjective information using natural language processing ( nlp )
0
this gives a principled mechanism to model hierarchical topic segmentation---in this paper , the latent topics are constrained to produce a hierarchical segmentation structure
1
as a strong baseline , we trained the skip-gram model of mikolov et al using the publicly available word2vec 5 software---as textual features , we use the pretrained google news word embeddings , obtained by training the skip-gram model with negative sampling
1
peters et al show how deep contextualized word representations model both complex characteristics of word use , and usage across various linguistic contexts---peters et al proposed the embeddings from language models , which obtains contextualized word representations
1
besides the msubased method , we use a substring tagging strategy to generate local substring tagging candidates---dinu and lapata and s茅aghdha and korhonen introduced a probabilistic model to represent word meanings by a latent variable model
0
we used 200 dimensional glove word representations , which were pre-trained on 6 billion tweets---for representing words , we used 100 dimensional pre-trained glove embeddings
1
although coreference resolution is a subproblem of natural language understanding , coreference resolution evaluation metrics have predominately been discussed in terms of abstract entities and hypothetical system errors---coreference resolution is the process of finding discourse entities ( markables ) referring to the same real-world entity or concept
1
twitter is a popular microblogging service which provides real-time information on events happening across the world---twitter is a widely used social networking service
1
the incorrectly predicted alignment types are shown with the ∗ symbol---type and the horizontal axis represents the predicted alignment type
1
the overall objective to minimize hence is math-p-3-8-0 where math-w-3-9-0-1 is the regularization strength---by minimizing a lifted loss math-w-7-1-0-10 , the tuple-embedding space needs to be restricted to math-w-7-1-0-22
1
adagrad with minibatch is adopted for optimization---adagrad with mini-batches is employed for optimization
1
such lists are usually composed of around 100 single terms---such lists are usually composed of about 100 single terms
1
its relative performance was 0.92 to 0.97 compared to directly trained smt systems---and that its relative performance compared to the directly trained smt systems was 0 . 92 to 0 . 97
1
the relation ] denotes the reflexive and transitive closure of ]---the relation > is the transitive closure of r
1
word sense disambiguation ( wsd ) is a task to identify the intended sense of a word based on its context---word sense disambiguation ( wsd ) is a natural language processing ( nlp ) task in which the correct meaning ( sense ) of a word in a given context is to be determined
1
score function is represented by math-w-3-1-0-67---that relation math-w-2-4-1-301 focuses on through the matrix
1
in this paper we extended the spectral learning ideas to learn a simple yet powerful dependency parser---neural networks , working on top of conventional n-gram back-off language models , have been introduced in as a potential means to improve conventional language models
0
in , non-expert annotators generated paraphrases for 250 noun-noun compounds , which were then used as the gold standard data for evaluating an automatic paraphrasing system---in workers generated paraphrases of 250 noun-noun compounds which were then used as the gold standard dataset for evaluating an automatic method of noun compound paraphrasing
1
marton and resnik took the source constituency tree into account and added soft constraints to the hierarchical phrasebased model---we apply dropout on the lstm layer to prevent network parameters from overfitting and control the co-adaptation of features
0
dependency parsing is a longstanding natural language processing task , with its outputs crucial to various downstream tasks including relation extraction ( cite-p-12-3-9 , cite-p-12-1-1 ) , language modeling ( cite-p-12-1-10 ) , and natural logic inference ( cite-p-12-1-4 )---we trained word embeddings using word2vec on 4 corpora of different sizes and types
0
our interest in this paper is the effect of alternative message wording , meaning how the message is said , rather than what the message is about---in this paper is the effect of alternative message wording , meaning how the message is said , rather than what the message is about
1
this is a well-known problem for bootstrapping approaches---the cotraining approach is well known for semi-supervised approach
1
cussens and pulman used a symbolic approach employing inductive logic programming , while erbach , barg and walther and fouvry followed a unificationbased approach---for instance , mihalcea et al studied pmi-ir , lsa , and six wordnet-based measures on the text similarity task
0
headden iii et al introduce the extended valence grammar and add lexicalization and smoothing---some researchers have applied the rule of transliteration to automatically translate proper names
0
a simple greedy algorithm is guaranteed to produce an approximately optimal summary---a greedy algorithm can obtain an approximately optimal summary
1
coreference resolution is the task of identifying all mentions which refer to the same entity in a document---coreference resolution is the next step on the way towards discourse understanding
1
table 2 shows the translation quality measured in terms of bleu metric with the original and universal tagset---table 1 presents the results from the automatic evaluation , in terms of bleu and nist test
1
the ud scheme is built on the google universal part-of-speech tagset , the interset interlingua of morphosyntactic features , and stanford dependencies---for training the trigger-based lexicon model , we apply the expectation-maximization algorithm
0
for all classifiers , we used the scikit-learn implementation---for this task , we used the svm implementation provided with the python scikit-learn module
1
some of the commonly used word representation techniques are word2vec , glove , neural language model , etc---there are various methods such word2vec and global vectors for word representation which create a distributed representation of words
1
coreference resolution is a multi-faceted task : humans resolve references by exploiting contextual and grammatical clues , as well as semantic information and world knowledge , so capturing each of these will be necessary for an automatic system to fully solve the problem---we used moses , a phrase-based smt toolkit , for training the translation model
0
for this task , we use the widely-used bleu metric---the smt weighting parameters were tuned by mert using the development data
0
the baseline of our approach is a statistical phrase-based system which is trained using moses---relation extraction ( re ) is a task of identifying typed relations between known entity mentions in a sentence
0
in live chats , wu et al and forsyth defined 15 dialogue acts for casual online conversations based on previous sets and characteristics of conversations---to calculate language model features , we train traditional n-gram language models with ngram lengths of four and five using the srilm toolkit
0
word sense disambiguation ( wsd ) is a key enabling technology that automatically chooses the intended sense of a word in context---the icsi meeting corpus consists of recordings of universitybased multi-speaker research meetings , totaling about 72 hours from 75 meetings
0
with reference to this system , we implement a data-driven parser with a neural classifier based on long short-term memory---for sequence modeling in all three components , we use the long short-term memory recurrent neural network
1
all word vectors are trained on the skipgram architecture---for each math-w-14-3-1-8 , define math-w-14-3-1-14
0
it has previously been shown that word embeddings represent the contextualised lexical semantics of words---it has been empirically shown that word embeddings can capture semantic and syntactic similarities between words
1
the toolkit enables the use of structural kernels in svm-light---researchers have developed framenet , a large lexical database of english that comes with sentences annotated with semantic frames
0
the base pcfg uses simplified categories of the stanford pcfg parser---the reordering rules are based on parse output produced by the stanford parser
1
in this work , we propose an alternative method to use amrs for abstractive summarization---in this work , we present a new method to do semantic abstractive summarization
1
in this example , the target word statements belongs to ( ¡°evokes¡± ) the frame s tatement---in this example , the target word statements belongs to ( ¡° evokes ¡± )
1
update summarization is a new challenge in multi-document summarization focusing on summarizing a set of recent documents relatively to another set of earlier documents---update summarization is a form of multi-document summarization where a document set must be summarized in the context of other documents assumed to be known
1
we showed that sopa is an extension of a one-layer cnn---we show that sopa is an extension of a one-layer cnn
1
in previous work , however , one of us attempted to characterize these differing properties in such a way that a single uniform architecture , appropriately parameterized , might be used for both natural language processes---in previous work , however , one of us attempted to characterize these differing properties in such a way that a single uniform architecture , appropriately parameterized , might be used for both natural-language processes
1
lexical simplification is a popular task in natural language processing and it was the topic of a successful semeval task in 2012 ( cite-p-14-1-9 )---lexical simplification is a specific case of lexical substitution where the complex words in a sentence are replaced with simpler words
1
in this paper we presented a new method to re-embed words from offthe-shelf embeddings based on manifold learning---in this paper , we re-embed pre-trained word embeddings with a stage of manifold learning
1
also , a number of semi-supervised word aligners have been proposed---the model weights were trained using the minimum error rate training algorithm
0
over a gold standard of semantic annotations and concepts that best capture their arguments , the method substantially outperforms three baseline methods---over a gold standard of semantic annotations and concepts that best capture their arguments , the method substantially outperforms three baselines , on average , computing concepts that are less than one step in the hierarchy
1
semantic role labeling ( srl ) is the task of automatically annotating the predicate-argument structure in a sentence with semantic roles---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
1
part-of-speech ( pos ) tagging is a fundamental language analysis task---part-of-speech ( pos ) tagging is a well studied problem in these fields
1
coreference resolution is the task of identifying all mentions which refer to the same entity in a document---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
0
relation extraction ( re ) is the task of extracting semantic relationships between entities in text---relation extraction is the task of automatically detecting occurrences of expressed relations between entities in a text and structuring the detected information in a tabularized form
1
these qlf constructs are removed by the processes of quantifier scoping and reference resolution ( see below )---which engenders a natural separation between the compositional semantics and the processes of scoping and reference resolution
1
in this paper , we made the simple observation that questions about images often contain premises implied by the question and that reasoning about premises can help vqa models respond more in-tion to an image , and select an appropriate path of action---in this paper , we make a simple observation that questions about images often contain premises ¨c objects and relationships implied by the question ¨c and that reasoning about premises can help visual question answering ( vqa )
1
chambers et al focused on classifying the temporal relation type of event-event pairs using previously learned event attributes as features---chambers et al , 2007 ) focused on event-event relations using previously learned event attributes
1