text
stringlengths
82
736
label
int64
0
1
we use sentiwordnet for introducing sentiment of a word---we also employ the general sentiment lexicons , sentiwordnet , to connect opinions
1
wellner et al used the graphbank , which contains 105 associated press and 30 wall street journal articles annotated with discourse relations---we present new conceptual tasks : visual paraphrasing ( § 5 ) , creative image captioning , and creative visual paraphrasing ( § 7 ) , interleaved with corresponding experimental results ( § 6 , § 8 )
0
sentiment analysis is the computational analysis of people ’ s feelings or beliefs expressed in texts such as emotions , opinions , attitudes , appraisals , etc . ( cite-p-11-3-3 )---sentiment analysis is the task of identifying positive and negative opinions , sentiments , emotions and attitudes expressed in text
1
we notice erratic behavior when optimizing sparse feature weights with m 2 and offer partial solutions---but we describe optimizer hyperparameters that make sparse feature tuning with m 2 feasible
1
in this paper we propose a supervised and a semi-supervised method to disambiguate partial cognates between two languages : french and english---we describe a supervised and also a semi-supervised method to discriminate the senses of partial cognates between french and english
1
in this paper we describe our submission to semeval-2018 task 1 : affects in tweets---in this paper we have described affecthor , the system which we submitted to the semeval-2018 affects in tweets
1
we used 14 datasets with non-projective dependencies from the conll-2006 and conll-2008 shared tasks---we used 14 datasets , most of which are non-projective , from the conll 2006 and 2008 shared tasks
1
the evaluation metric for the overall translation quality is caseinsensitive bleu4---context have not been systematically compared for different word embeddings
0
but in a web crawl , the distribution is quite likely to be more uniform , which means the senses will ¡°split the difference¡± in the representation and end up not being that similar to any instance of serve---for our baseline we use the moses software to train a phrase based machine translation model
0
kilicoglu and bergler apply a combination of lexical and syntactic methods , improving on previous results and showing that quantifying the strength of a hedge can be beneficial for classification of speculative sentences---kilicoglu and bergler apply a linguistically motivated approach to the same classification task by using knowledge from existing lexical resources and incorporating syntactic patterns
1
the system used a tri-gram language model built from sri toolkit with modified kneser-ney interpolation smoothing technique---the language model used was a 5-gram with modified kneserney smoothing , built with srilm toolkit
1
bilingual word embeddings has become a source of great interest in recent times---headden , johnson and mcclosky introduced the extended valence grammar and added lexicalization and smoothing
0
sentiment analysis is a fundamental problem aiming to give a machine the ability to understand the emotions and opinions expressed in a written text---sentiment analysis is the task of automatically identifying the valence or polarity of a piece of text
1
the dominant approach to word alignment has been the ibm models together with the hmm model---ibm models and the hidden markov model for word alignment are the most influential statistical word alignment models
1
we use the logistic regression classifier in the skll package , which is based on scikit-learn , optimizing for f 1 score---for training the model , we use the linear kernel svm implemented in the scikit-learn toolkit
1
these nlp tools have the potential to make a marked difference for gun violence researchers---nlp researchers have the potential to significantly advance gun violence research
1
table 4 shows labeled and unlabeled accuracy scores of previous work reported for the penn2malt conversion with the head finding rules of yamada and matsumoto---table 3 gives the results for the penn treebank converted with the head-finding rules of yamada and matsumoto and the labeling rules of nivre
1
we extended the unsupervised corpus-extracted phrase approximation method of guevara and baroni and zamparelli to estimate all known state-of-the-art cdsms , using closedform solutions or simple iterative procedures in all cases---recently , neural networks based methods are proposed to learn the distributed representation of words on large scale of corpus
0
recently , deep reinforcement learning has attracted growing attention in the field of visual captioning---zarrie脽 and kuhn argue that multiword expressions can be reliably detected in parallel corpora by using dependency-parsed , word-aligned sentences
0
this paper proposes a novel method of jointly embedding knowledge graphs and logical rules---word sense disambiguation ( wsd ) is a fundamental task and long-standing challenge in natural language processing ( nlp )
0
our word embeddings is initialized with 100-dimensional glove word embeddings---we use pre-trained 50-dimensional word embeddings vector from glove
1
coreference resolution is the process of linking together multiple expressions of a given entity---coreference resolution is the next step on the way towards discourse understanding
1
a low-rank approximation of the tensor is then derived using a tensor decomposition---mapping is derived through tensor decomposition , which provides a low-rank approximation of the original tensor
1
we use the publicly available 300-dimensional word vectors of mikolov et al , trained on part of the google news dataset---we use distributed word vectors trained on the wikipedia corpus using the word2vec algorithm
1
readability is used to provide users with high-quality service in text recommendation or text visualization---readability is used to provide documents to non-expert users so that they can read the retrieved documents easily
1
for annotation tasks , snow et al showed that crowdsourced annotations are similar to traditional annotations made by experts---snow et al applied crowdsourcing to five nlp annotation tasks , but the settings of these tasks are very simple
1
figure 5 : examples of asia ’ s input and output---figure 5 shows some real examples of asia ’ s input and output
1
coreference resolution is a well known clustering task in natural language processing---coreference resolution is the problem of identifying which mentions ( i.e. , noun phrases ) refer to which real-world entities
1
a major challenge facing this task is the system coverage , i.e. , for any user-created nonstandard term , the system should be able to restore the correct word within its top n output candidates---with ¡° broad coverage ¡± , i . e . , for any user-created nonstandard token , the system should be able to restore the correct word within its top
1
our cdsm feature is based on word vectors derived using a skip-gram model---all word vectors are trained on the skipgram architecture
1
it is a standard phrasebased smt system built using the moses toolkit---the baseline system is a pbsmt engine built using moses with the default configuration
1
cogenthelp is a prototype tool for authoring dynamicallygenerated online help for applications with graphical user interfaces , embodying the evolution-friendly properties of tools in the literate programming tradition---cogenthelp is a prototype tool for authoring dynamically generated online help for applications with graphical user interfaces ( guis )
1
word sense disambiguation ( wsd ) is a key enabling-technology that automatically chooses the intended sense of a word in context---word sense disambiguation ( wsd ) is the task of determining the meaning of a word in a given context
1
we use 300-dimensional glove vectors trained on 6b common crawl corpus as word embeddings , setting the embeddings of outof-vocabulary words to zero---domain adaptation techniques have been employed in nlp
0
in addition , a 5-gram lm with kneser-ney smoothing and interpolation was built using the srilm toolkit---in this paper we present s up wsd , whose objective is to overcome the aforementioned drawbacks , and facilitate the use of a supervised wsd software
0
we used the srilm software 4 to build langauge models as well as to calculate cross-entropy based features---we used the srilm toolkit to simulate the behavior of flexgram models by using count files as input
1
in contrast , our approach is designed to acquire temporal relations across sentences in a narrative paragraph---mimno et al extend the original concept of lda to support polylingual topic models , both on parallel and partly comparable documents
0
in principle , the cache-based approach can be well suited for document-level translation---we use a minibatch stochastic gradient descent algorithm together with an adagrad optimizer
0
learning from query logs also allows us to leverage the concept of user intents---user intents can be an important factor in modeling type
1
we show empirically that , although adding metadata improves the performance on standard metrics , it favors self-citations which are less useful in a citation recommendation setup---on standard metrics , we found that it introduces a bias for self-citation which might not be desirable in a citation recommendation system
1
coreference resolution is a task aimed at identifying phrases ( mentions ) referring to the same entity---automatically solving math word problems has proved a difficult and interesting challenge for the ai research community
0
in this paper , we propose an adaptive ensemble method to adapt coreference resolution across domains---in this paper , we proposed an adaptive ensemble method for coreference resolution
1
we computed the translation accuracies using two metrics , bleu score , and lexical accuracy on a test set of 30 sentences---we used the sri language modeling toolkit to train a fivegram model with modified kneser-ney smoothing
0
we aim to capture word reordering knowledge for the attention-based nmt by incorporating distortion models---word reordering knowledge needs to be incorporated into attention-based nmt
1
we use bnc and a list of verb-noun constructions extracted from bnc by fazly et al and cook et al and labeled as l , i , or q---we use bnc and a list of verbnoun constructions extracted from bnc by fazly et al , cook et al , i , or q
1
the scikit-learn library was used for the svm , which utilized a polynomial kernel with degree of 4---the scikit-learn implementation of the svc-class with a linear kernel was used
1
mikolov et al presents a neural network-based architecture which learns a word representation by learning to predict its context words---mikolov et al proposed vector representation of words with the help of negative sampling that improves both word vector quality and training speed
1
overall , our experiments show that current vqa attention models do not seem to be looking at the same regions as humans---vqa-attention maps remain the same , which confirms our key finding that current vqa attention models do not seem to be looking at the same regions as humans
1
xia et al automatically extracted conversion rules from a target treebank and proposed strategies to handle the case when more than one conversion rule are applicable---we use glove vectors with 200 dimensions as pre-trained word embeddings , which are tuned during training
0
we used moses as the phrase-based machine translation system---we used moses , a phrase-based smt toolkit , for training the translation model
1
in this work we have illustrated the need for incorporating world knowledge in training task specific models---in this work , we propose to enhance learning models with world knowledge in the form of knowledge
1
besides , chinese is a topic-prominent language , the subject is usually covert and the usage of words is relatively flexible---in recent years , neural network models have been introduced to n-er task
0
the feature weights 位 m are tuned with minimum error rate training---the feature weights are tuned with mert to maximize bleu-4
1
we evaluated the translation quality using the bleu-4 metric---our hypothesis is a generalization of the original hypothesis since it allows a reducible sequence to form several adjacent subtrees
0
text simplification essentially is the process of rewriting a given text to make it easier to process for a given audience---the translation quality is evaluated by bleu and ribes
0
an effective solution for these problems is the long short-term memory architecture---long short-term memory have been proposed as a solution for the rnns issue , introducing a memory cell inside the network
1
the decoding weights are optimized with minimum error rate training to maximize bleu scores---the model weights are automatically tuned using minimum error rate training
1
sentiment analysis in twitter is a particularly challenging task , because of the informal and “ creative ” writing style , with improper use of grammar , figurative language , misspellings and slang---in this paper , we study the differences among sms normalization , general text normalization , spelling
0
we used a 5-gram language model with modified kneser-ney smoothing implemented using the srilm toolkit---we propose a probabilistic approach for performing joint query annotation
0
the log linear weights for the baseline systems are optimized using mert provided in the moses toolkit---the log-linear parameter weights are tuned with mert on the development set
1
culotta and sorensen described a slightly generalized version of this kernel based on dependency trees---stance detection is the task of automatically determining from text whether the author of the text is in favor of , against , or neutral towards a proposition or target
0
fader et al learned question paraphrases from aligning multiple questions with the same answers generated by wikianswers---wikianswers fader et al extracted the similar questions on wikianswers and used them as question paraphrases
1
the b & b and m ar m o t models are single-source---b & b and m ar m o t models are single-source
1
results indicate that integration of situational context dramatically improves performance over traditional methods alone---tests show that using a situated model significantly improves performances over traditional language modeling methods
1
for both languages , we used the srilm toolkit to train a 5-gram language model using all monolingual data provided---we trained a 5-gram language model on the xinhua portion of gigaword corpus using the srilm toolkit
1
in this paper , we propose to use word predictions as a mechanism for direct supervision---we propose to use the word prediction mechanism to enhance the initial state generated by the encoder
1
extensive experiments have validated the effectiveness of the corpus-based method for classifying the word ’ s sentiment polarity---extensive experiments have validated the effectiveness of the corpus-based method in polarity classification task
1
further , we apply a 4-gram language model trained with the srilm toolkit on the target side of the training corpus---we used the mstparser as the basic dependency parsing model
0
for word-level embedding e w , we utilize pre-trained , 300-dimensional embedding vectors from glove 6b---for the first two features , we adopt a set of pre-trained word embedding , known as global vectors for word representation
1
we created 5-gram language models for every domain using srilm with improved kneserney smoothing on the target side of the training parallel corpora---we trained a 4-gram language model with kneser-ney smoothing and unigram caching using the sri-lm toolkit
1
word sense disambiguation ( wsd ) is the task to identify the intended sense of a word in a computational manner based on the context in which it appears ( cite-p-13-3-4 )---we use the svm implementation available in the li-blinear package
0
rhetorical structure theory is a framework for describing the organization of a text and what a text conveys by identifying hierarchical structures in text---in this section we relate our work with the existing literature
0
we use the penn discourse treebank , which is the largest handannotated discourse relation corpus annotated on 2312 wall street journal articles---in this paper , we focus on the problem of using sentence compression techniques
0
with the svm reranker , we obtain a significant improvement in bleu scores over white & rajkumar ’ s averaged perceptron model on both development and test data---neg-finder successfully removes the necessity of including manually crafted supervised knowledge
0
we use minimal error rate training to maximize bleu on the complete development data---we propose an unsupervised model that identifies recap segments
0
word embeddings have recently gained popularity among natural language processing community---the use of unsupervised word embeddings in various natural language processing tasks has received much attention
1
we implemented our method in a phrase-based smt system---we used moses as the implementation of the baseline smt systems
1
we found that performance improves steadily as the number of available languages increases---we ’ ve demonstrated that the benefits of unsupervised multilingual learning increase steadily with the number of available languages
1
experiments on large scale real-life ¡°yahoo ! answers¡± dataset reveals that scqa outperforms current state-of-the-art approaches based on translation models , topic models and deep neural netwo---experiments on large scale real-life ¡° yahoo ! answers ¡± dataset revealed that t-scqa outperforms current state-of-the-art approaches
1
word embedding we use the word2vec toolkit to pre-train word embeddings on the whole english wikipedia dump---we preinitialize the word embeddings by running the word2vec tool on the english wikipedia dump
1
for both languages , we used the srilm toolkit to train a 5-gram language model using all monolingual data provided---we use srilm train a 5-gram language model on the xinhua portion of the english gigaword corpus 5th edition with modified kneser-ney discounting
1
the embedding layer was initialized using word2vec vectors---modified kneser-ney trigram models are trained using srilm on the chinese portion of the training data
0
word subject domains have been widely used to improve the performance of word sense disambiguation algorithms---word subject domains have been widely used to improve the performance of machine translation systems
1
system tuning was carried out using minimum error rate training optimised with k-best mira on a held out development set---system tuning was carried out using both k-best mira and minimum error rate training on the held-out development set
1
the irstlm toolkit is used to build language models , which are scored using kenlm in the decoding process---we trained a 4-gram language model with kneser-ney smoothing and unigram caching using the sri-lm toolkit
0
to the best of our knowledge , this is the first time that very deep convolutional nets have been applied to text processing---we use opinionfinder which employs negative and positive polarity cues
0
the language model is a 5-gram with interpolation and kneserney smoothing---this type of features are based on a trigram model with kneser-ney smoothing
1
with experiments on many relations from two separate knowledge bases , we have shown that our methods significantly outperform prior work on knowledge base inference---we first use bleu score to perform automatic evaluation
0
a 4-gram language model is trained on the xinhua portion of the gigaword corpus with the srilm toolkit---a 4-gram language model was trained on the target side of the parallel data using the srilm toolkit from stolcke
1
sentiment analysis is a natural language processing task whose aim is to classify documents according to the opinion ( polarity ) they express on a given subject ( cite-p-13-8-14 )---sentiment analysis is the task of identifying positive and negative opinions , sentiments , emotions and attitudes expressed in text
1
in this paper , we illustrate such importance using named entity ( ne ) translation mining problem---in this paper , we explore latent features of temporality for understanding relation
1
in the most likely scenario – porting a parser to a novel domain for which there is little or no annotated data – the improvements can be quite large---as stated above , we aim to build an unsupervised generative model for named entity clustering , since such a model could be integrated with unsupervised coreference models like haghighi and klein for joint inference
0
we used a 5-gram language model with modified kneser-ney smoothing implemented using the srilm toolkit---the srilm toolkit was used to build the trigram mkn smoothed language model
1
we use the publicly available 300-dimensional word vectors of mikolov et al , trained on part of the google news dataset---we use the 300-dimensional skip-gram word embeddings built on the google-news corpus
1
these results support the use of heterogeneous measures in order to consolidate text evaluation results---that suggest the convenience of using heterogeneous measures to corroborate evaluation results
1
in this paper , we describe a probabilistic answer ranking framework for multiple languages---in this paper , we presented a generalized answer selection framework which was applied to chinese and japanese
1
word sense disambiguation ( wsd ) is the task of determining the correct meaning or sense of a word in context---word sense disambiguation ( wsd ) is a fundamental task and long-standing challenge in natural language processing ( nlp )
1
first , we train a vector space representations of words using word2vec on chinese wikipedia---for our purpose we use word2vec embeddings trained on a google news dataset and find the pairwise cosine distances for all words
1
we measure translation quality via the bleu score---we measure the translation quality using a single reference bleu
1
the algorithms were implemented using scikit-learn , a general purpose machine learning python library---latent dirichlet allocation is one of the most popular topic models used to mine large text data sets
0