sentence1
stringlengths 16
446
| sentence2
stringlengths 14
436
|
|---|---|
the feature weights are tuned to optimize bleu using the minimum error rate training algorithm . | the nnlm weights are optimized as the other feature weights using minimum error rate training . |
additionally , we compare our system to the finnish data-driven morphological analyzer presented by silfverberg and hulden . | in our second experiment , we compare our system against the neural morphological analyzer proposed by silfverberg and hulden . |
distributed word embeddings are learned using a skip-gram recurrent neural net architecture running over a large raw corpus . | these word embeddings are learned in advance using a continuous skip-gram model , or other continuous word representation learning methods . |
modified kneser-ney trigram models are trained using srilm on the chinese portion of the training data . | gram language models are trained over the target-side of the training data , using srilm with modified kneser-ney discounting . |
we proposed two new voting methods according to the characteristics of the chunking task . | 2 ) we propose two novel voting methods based on the characteristics of chunking task . |
we compare inquiry semantics to other kinds of semantics , and also identify the nature of meaning . | finally we characterize inquiry semantics and the notion of meaning . |
transliteration is the task of converting a word from one alphabetic script to another . | transliteration is often defined as phonetic translation ( cite-p-21-3-2 ) . |
in this paper , we propose a joint learning method of two smt systems for paraphrase generation . | in this paper , we propose a joint learning method of two smt systems to optimize the process of paraphrase generation . |
we use the word2vec tool to pre-train the word embeddings . | we use the skip-gram model , trained to predict context tags for each word . |
for the turn , the idtb model enables negotiative turn-taking and supports true mixed-initiative interaction . | in short , the importance-driven turnbidding model provides a negotiative turn-taking framework that supports mixed-initiative interactions . |
we implement classification models using keras and scikit-learn . | we implement logistic regression with scikit-learn and use the lbfgs solver . |
shallow semantic representations can prevent the sparseness of deep structural approaches and the weakness of cosine similarity based models . | shallow semantic representations could prevent the sparseness of deep structural approaches and the weakness of bow models . |
metaphor is a type of analogy and is closely related to other rhetorical figures of speech that achieve their effects via association , comparison or resemblance including allegory , hyperbole , and simile . | metaphor is a common linguistic tool in communication , making its detection in discourse a crucial task for natural language understanding . |
neural networks ( rnns ) are one of the most prevalent architectures because of the ability to handle variable-length texts . | recently , neural networks have achieved great success on sentiment classification due to their ability to alleviate feature engineering . |
lda is a probabilistic model that can be used to model and discover underlying topic structures of documents . | lda is a generative probabilistic model where documents are viewed as mixtures over underlying topics , and each topic is a distribution over words . |
coreference resolution is a key problem in natural language understanding that still escapes reliable solutions . | additionally , coreference resolution is a pervasive problem in nlp and many nlp applications could benefit from an effective coreference resolver that can be easily configured and customized . |
the results evaluated by bleu score is shown in table 2 . | the scores of participants are in table 10 in terms of bleu and f 1 scores . |
twitter is a well-known social network service that allows users to post short 140 character status update which is called β tweet β . | twitter is a subject of interest among researchers in behavioral studies investigating how people react to different events , topics , etc. , as well as among users hoping to forge stronger and more meaningful connections with their audience through social media . |
we use scikit learn python machine learning library for implementing these models . | we implemented linear models with the scikit learn package . |
among these techniques , latent semantic indexing is a wellknown approach . | the most commonly used and prominent ones are latent semantic indexing and probabilistic lsi . |
our method involved using the machine translation software moses . | for training the translation model and for decoding we used the moses toolkit . |
in this paper , we present the lth coreference solver used in the closed track of the conll 2012 shared task . | in this paper we present our contribution to the conll 2012 shared task . |
hence , we introduce an attention mechanism to extract the words that are important to the meaning of the post , and aggregate the representation of those informative words to form a vector . | in our model , we use an attention mechanism to integrate the information from a set of comment into an action embedding vector . |
experimental results demonstrate that our proposed method outperforms three kb-qa baseline methods . | the experimental results show that our method outperforms the latest three kbqa baseline systems . |
experimental results on the japanese-english language pair show a relative error reduction of 4 % of the alignment score compared to a model with 1-best parse trees that using forest . | experimental results show a relative error reduction of 4 % of the alignment score compared to the model with 1-best parse trees . |
a major component in phrase-based statistical machine translation is the table of conditional probabilities of phrase translation pairs . | standard phrase-based machine translation uses relative frequencies of phrase pairs to estimate a translation model . |
in recent years has created an increasing need for improvements in organic and sponsored search . | growth and competition in web search in recent years has created an increasing need for improvements in organic and sponsored search . |
a metaphor is a literary figure of speech that describes a subject by asserting that it is , on some point of comparison , the same as another otherwise unrelated object . | a metaphor is a figure of speech that creates an analogical mapping between two conceptual domains so that the terminology of one ( source ) domain can be used to describe situations and objects in the other ( target ) domain . |
generation of referring expression is an important task in the field of natural language generation systems . | referring expression generation is an important problem in natural language generation . |
one distance measure for urll trees is introduced in . | one such urll distance measure is given in . |
several systems that can accommodate non-projective structures have subsequently been described . | extensions for transition systems have been proposed to handle non-projective structures with additional actions . |
gildea and jurafsky applied sp to automatic srl by clustering extracted verb-direct object pairs , resulting in modest improvements . | gildea and jurafsky presented an early framenet-based srl system that targeted both verbal and nominal predicates . |
the graph formalization that underlies autoextend is based on the offset calculus introduced by mikolov et al . | our multi-modal architecture builds on the continuous log-linear skipgram language model proposed by mikolov et al . |
coherence is a common 'currency ' with which to measure the benefit of applying a schema . | since coherence is a measure of how much sense the text makes , it is a semantic property of the text . |
we evaluate our approaches on eight language pairs , with training data sizes ranging from 100k words to 8m words , and show improvements of up to + 4 . 3 bleu , surpassing phrase-based translation in nearly all settings . | we evaluate our approaches on eight language pairs with data sizes ranging from 100k to 8m words , and achieve improvements of up to +4.3 bleu , surpassing phrase-based translation in nearly all settings . |
as neural network based models dominate the research in natural language processing , seq2seq models have been widely used for response generation . | recurrent neural network architectures have proven to be well suited for many natural language generation tasks . |
we used the srilm toolkit to create 5-gram language models with interpolated modified kneser-ney discounting . | however , we use a large 4-gram lm with modified kneser-ney smoothing , trained with the srilm toolkit , stolcke , 2002 and ldc english gigaword corpora . |
relation extraction ( re ) is the task of extracting instances of semantic relations between entities in unstructured data such as natural language text . | relation extraction is the task of detecting and characterizing semantic relations between entities from free text . |
burstein et al employ this idea for evaluating coherence in student essays . | burstein et al use the entity-grid for student essay evaluation , which is a scenario closer to ours . |
the model is a log-linear model over synchronous cfg derivations . | the whole translation model is organized in a log-linear framework . |
lexical simplification is the task to find and substitute a complex word or phrase in a sentence with its simpler synonymous expression . | lexical simplification is the task of modifying the lexical content of complex sentences in order to make them simpler . |
using these contexts without smoothing leads to data sparsity problems ; therefore we have developed decision tree clustering algorithms to cluster source word contexts based on optimisation of the em auxiliary function . | since the use of these contexts alone causes data sparsity problems , we develop a decision tree algorithm for clustering the contexts based on optimisation of the em auxiliary function . |
dependency parsing is a central nlp task . | therefore , dependency parsing is a potential β sweet spot β that deserves investigation . |
we used trigram language models with interpolated kneser-kney discounting trained using the sri language modeling toolkit . | we further used a 5-gram language model trained using the srilm toolkit with modified kneser-ney smoothing . |
su et al presented a clustering method that utilizes the mutual reinforcement associations between features and opinion words . | su et al used a clustering method to map the implicit aspect candidates to explicit aspects . |
for english , we use the stanford parser for both pos tagging and cfg parsing . | we use the stanford parser for english language data . |
nguyen et al described methods of detecting and correcting ws inconsistencies in the vtb corpus . | nguyen et al analyzed n-gram sequences and phrase structures to detect ws inconsistencies . |
the skip-gram model aims to find word representations that are useful for predicting the surrounding words in a sentence or document . | the word representation description w2v word2vec uses the skip-ngram model to find word representations that are useful to predict the surrounding words of a sentence or a document . |
we use the opensource moses toolkit to build a phrase-based smt system . | our implementation of the segment-based imt protocol is based on the moses toolkit . |
semantic role labeling ( srl ) is a major nlp task , providing a shallow sentence-level semantic analysis . | semantic role labeling ( srl ) is the task of automatically annotating the predicate-argument structure in a sentence with semantic roles . |
but the published method for calculating the entropy gradient requires significantly more computation than supervised crf training . | however , the method they present for calculating the gradient of the entropy takes substantially greater time than the traditional supervised-only gradient . |
the most commonly used word embeddings were word2vec and glove . | commonly used word vectors are word2vec , glove and fasttext . |
dependency parsing is a topic that has engendered increasing interest in recent years . | dependency parsing is a way of structurally analyzing a sentence from the viewpoint of modification . |
a 5-gram language model of the target language was trained using kenlm . | the smt systems used a kenlm 5-gram language model , trained on the mono-lingual data from wmt 2015 . |
xing et al incorporate length normalization in the training of word embeddings and try to maximize the cosine similarity instead , introducing an orthogonality constraint to preserve the length normalization after the projection . | xing et al incorporate length normalization in the training of word embeddings and maximize the cosine similarity instead , enforcing the orthogonality constraint to preserve the length normalization after the mapping . |
mln has been applied in several natural language processing tasks and demonstrated its advantages . | mln framework has been adopted for several natural language processing tasks and achieved a certain level of success . |
across various settings , this adversarial learning mechanism can significantly improve the performance of some of the most commonly used translation based kge methods . | experimental results show that adversarial training substantially improves the performances of target embedding models under various settings . |
roth and yih use ilp to deal with the joint inference problem of named entity and relation identification . | roth and yih develop a relation extraction approach that exploits constraints among entity types and the relations allowed among them . |
neural models for natural language generation based on the encoder-decoder framework have become quite popular recently . | recurrent neural network architectures have proven to be well suited for many natural language generation tasks . |
for language model scoring , we use the srilm toolkit training a 5-gram language model for english . | we train a 4-gram language model on the xinhua portion of english gigaword corpus by srilm toolkit . |
grenager and manning address the role induction problem and propose a directed graphical model which relates a verb , its semantic roles , and their possible syntactic realizations . | grenager and manning 1118 propose a directed graphical model which relates a verb , its semantic roles , and their possible syntactic realizations . |
we analyze the problem of joint models on the task of ed , and propose to use the annotated argument information explicitly for this task . | in this work , we propose to exploit argument information explicitly for ed via supervised attention mechanisms . |
for word-level embedding e w , we utilize pre-trained , 300-dimensional embedding vectors from glove 6b . | we initialize the word embedding matrix with pre-trained glove embeddings . |
twitter is a widely used microblogging environment which serves as a medium to share opinions on various events and products . | twitter is a popular microblogging service which provides real-time information on events happening across the world . |
in the above examples , classifier β hiki β is used to count noun β inu ( dog ) β . | in the above examples , classifier β hiki β is used to count noun β inu ( dog ) β , while β satsu β for β hon ( book ) β . |
misra et al use a latent dirichlet allocation topic model to find coherent segment boundaries . | li et al used a latent dirichlet allocation model to generate topic distribution features as the news representations . |
we use 50 dimensional word embeddings , which are initialized by the 50 dimensional pre-trained word vectors 6 from glove , and updated in the training process . | we use 100-dimension glove vectors which are pre-trained on a large twitter corpus and fine-tuned during training . |
hatzivassiloglou and mckeown used a log-linear regression model to predict the similarity of conjoined adjectives . | hatzivassiloglou and mckeown showed how the pattern x and y could be used to automatically classify adjectives as having positive or negative orientation . |
a small number of documents may indicate that the annotated data provide a limited coverage of various lexical and semantic phenomena . | as a result , only a small number of documents are typically annotated , limiting the coverage of various lexical/semantic phenomena . |
kay proposes a framework with which each of the autosegmental tiers is assigned a tape in a multi-tape finite state machine , with an additional tape for the surface form . | kay proposes a framework for handling templatic morphology in which each templatic morpheme is assigned a tape in a multi-tape finite state machine , with an additional tape for the surface form . |
sentiment analysis ( sa ) is a field of knowledge which deals with the analysis of people β s opinions , sentiments , evaluations , appraisals , attitudes and emotions towards particular entities ( liu , 2012 ) . | sentiment analysis ( sa ) is the research field that is concerned with identifying opinions in text and classifying them as positive , negative or neutral . |
the first study was conducted by ferretti et al , who found that verbs facilitate the processing of nouns denoting prototypical participants in the depicted event and of adjectives denoting features of prototypical participants . | the first dataset comes from ferretti et al , who found that verbs facilitate the processing of nouns denoting prototypical participants in the depicted event and of adjectives denoting features of prototypical participants . |
we use existing , freely-available clusters trained on news data by turian et al using the implementation by liang . | regarding brown clusters , we use freely available clusters trained on news data by turian et al using the implementation by liang . |
we used the logistic regression implementation in scikit-learn for the maximum entropy models in our experiments . | we used the logistic regression implemented in the scikit-learn library with the default settings . |
in an early experiment , cite-p-17-4-21 analyzed the acoustic properties of the / d / sound . | in an early experiment , cite-p-17-4-21 analyzed the acoustic properties of the /d/ sound in the two syllables /di/ and /du/ . |
we use stochastic gradient descent with adagrad , l 2 regularization and minibatch training . | parameter optimization is performed with the diagonal variant of adagrad with minibatchs . |
recently , question generation has got immense attention from the researchers and hence , different methods have been proposed to accomplish the task in different relevant fields . | recently , question generation has received immense attention from researchers and different methods have been proposed to accomplish the task in different relevant fields . |
the texts were pos-tagged , using the same tag set as in the penn treebank . | the pos tags are based on penn treebank pos tagset . |
a multiword expression is any combination of words with lexical , syntactic or semantic idiosyncrasy , in that the properties of the mwe are not predictable from the component words . | more generally , collocations are a frequent type of multiword expression , a sequence of words that presents some lexical , syntactic , semantic , pragmatic or statistical idiosyncrasies . |
coreference resolution is the task of determining which mentions in a text are used to refer to the same real-world entity . | coreference resolution is a task aimed at identifying phrases ( mentions ) referring to the same entity . |
by exploiting semantic similarities between dialogue utterances and ontology terms , the model alleviates the need for ontology-dependent parameters . | in this paper , a novel approach is introduced that fully utilizes semantic similarity between dialogue utterances and the ontology terms , allowing the information to be shared across domains . |
as abney shows , we can not use relatively simple techniques such as relative frequencies to obtain a model for estimating derivation probabilities in attribute-value grammars . | however abney showed that attribute-value grammars can not be modeled adequately using statistical techniques which assume that statistical dependencies are accidental . |
starting with this graph , we use the graph iteration algorithm from to calculate a score for each vertex in the graph . | by employing the graph iteration algorithm proposed in , we can compute the rank of a vertex in the entire graph . |
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 . | we use sri language model toolkit to train a 5-gram model with modified kneser-ney smoothing on the target-side training corpus . |
in this paper , i describe how donnellan ' s distinction between referential and attributive . | in this paper , i present a general framework for treating donnellan 's distinction . |
we used a phrase-based smt model as implemented in the moses toolkit . | for all submissions , we used the phrase-based variant of the moses decoder . |
pereira et al use an information-theoretic based clustering approach , clustering nouns according to their distribution as direct objects among verbs . | pereira et al cluster nouns according to their distribution as direct objects of verbs , using information-theoretic tools . |
additionally letting our model learn the language β s canonical word order improves its performance and leads to the highest semantic parsing . | finally , we explain how to let our model additionally learn the language β s canonical word order . |
for the neural models , we use 100-dimensional glove embeddings , pre-trained on wikipedia and gigaword . | we use 300-dimensional glove vectors trained on 6b common crawl corpus as word embeddings , setting the embeddings of outof-vocabulary words to zero . |
djuric et al propose an approach that learns low-dimensional , distributed representations of user comments in order to detect expressions of hate speech . | djuric et al propose to learn distributed lowdimensional representations of comments in order to use them as a feature for logistic regression . |
wordnet is a key lexical resource for natural language applications . | unfortunately , wordnet is a fine-grained resource , encoding sense distinctions that are difficult to recognize even for human annotators ( cite-p-13-1-2 ) . |
the language model was trained using srilm toolkit . | the language models were trained using srilm toolkit . |
learningbased approaches were first applied to identify within-sentence discourse relations , and only later to cross-sentence text-level relations . | learning-based approaches were first applied to identify within-sentence discourse relations , and only later to cross-sentence relations at the document level . |
modified kneser-ney trigram models are trained using srilm on the chinese portion of the training data . | modified kneser-ney trigram models are trained using srilm upon the chinese portion of the training data . |
the language model is a 3-gram language model trained using the srilm toolkit on the english side of the training data . | the lm uses the monolingual data and is trained as a five-gram 9 using the srilm-toolkit . |
the weights δ½ m in the log-linear model were trained using minimum error rate training with the news 2009 development set . | the log-linear feature weights are tuned with minimum error rate training on bleu . |
for the classification task , we use pre-trained glove embedding vectors as lexical features . | for the word-embedding based classifier , we use the glove pre-trained word embeddings . |
we have identified important issues encountered in using inference rules for textual entailment . | intuitively such inference rules should be effective for recognizing textual entailment . |
the skip-gram model adopts a neural network structure to derive the distributed representation of words from textual corpus . | the skip-gram model implemented by word2vec learns vectors by predicting context words from targets . |
we also explore bi-lstm models to avoid the detailed feature engineering . | the bi-lstm models reduce the cost of feature engineering . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.