NAACL 2016 论文列表

When translating into a morphologically complex language, segmenting the target language can reduce data sparsity, while introducing the complication of desegmenting the system output. We present a method for decoder-integrated desegmentation, allowing features that consider the desegmented target, such as a word-level language model, to be considered throughout the entire search space. Our re-sults on a large-scale, English to Arabic translation task show significant improvement over the 1-best desegmentation baseline.

Word clusters are useful for many NLP tasks including training neural network language models, but current increases in datasets are outpacing the ability of word clusterers to handle them. Little attention has been paid thus far on inducing high-quality word clusters at a large scale. The predictive exchange algo-rithm is quite scalable, but sometimes does not provide as good perplexity as other slower clustering algorithms. We introduce the bidirectional, interpolated, refining, and alternating (BIRA) predictive exchange algorithm. It improves upon the predictive exchange algorithm’s perplexity by up to 18%, giving it perplexities comparable to the slower two-sided exchange algorithm, and better perplexities than the slower Brown clustering algorithm. Our B IRA implementation is fast, clustering a 2.5 billion token English News Crawl corpus in 3 hours. It also reduces machine translation training time while preserving translation quality. Our implementation is portable and freely available.

We present Emergent, a novel data-set derived from a digital journalism project for rumour debunking. The data-set contains 300 rumoured claims and 2,595 associated news articles, collected and labelled by journalists with an estimation of their veracity (true, false or unverified). Each associated article is summarized into a headline and labelled to indicate whether its stance is for, against, or observing the claim, where observing indicates that the article merely repeats the claim. Thus, Emergent provides a real-world data source for a variety of natural language processing tasks in the context of fact-checking. Further to presenting the dataset, we address the task of determining the article headline stance with respect to the claim. For this purpose we use a logistic regression classifier and develop features that examine the headline and its agreement with the claim. The accuracy achieved was 73% which is 26% higher than the one achieved by the Excitement Open Platform (Magnini et al., 2014).

The goal of Event extraction is to extract structured information of events that are of interest from unstructured documents. Existing event extractors for social media suffer from two major problems: lack of context and informal nature. In this paper, instead of conducting event extraction solely on each social media message, we incorporate cross-genre knowledge to boost the event extractor performance. Experiment results demonstrate that without any additional annotations, our proposed approach is able to provide 15% absolute F-score improvement over the state-of-the-art.

Recent work across several AI subdisciplines has focused on automatically solving math word problems. In this paper we introduce MAWPS, an online repository of Math Word Problems, to provide a unified testbed to evaluate different algorithms. MAWPS allows for the automatic construction of datasets with particular characteristics, providing tools for tuning the lexical and template overlap of a dataset as well as for filtering ungrammatical problems from web-sourced corpora. The online nature of this repository facilitates easy community contribution. At present, we have amassed 3,320 problems, including the full datasets used in several prominent works.

This paper proposes an incremental learning strategy for neural word embedding methods, such as SkipGrams and Global Vectors. Since our method iteratively generates embedding vectors one dimension at a time, obtained vectors equip a unique property. Namely, any right-truncated vector matches the solution of the corresponding lower-dimensional embedding. Therefore, a single embedding vector can manage a wide range of dimensional requirements imposed by many different uses and applications.

Quotations are kernels not just of wisdom but also of beautiful and striking language. While recent studies have characterized the stylistic features of quotations, we characterize the order of stylistic information within quotations. Analyzing a corpus of two-sentence quotations collected from the social network Tum-blr, we explore the ways that both low-level features and high-level features tend to occur in either the first or second sentence. Through analysis of examples, we interpret these ten-dencies as manifestations of rhetorical patterns. Results from a prediction task suggest that stylistic patterns are more prominent in quotations than in a comparison corpus.

Research on grammatical error correction has received considerable attention. For dealing with all types of errors, grammatical error correction methods that employ statistical machine translation (SMT) have been proposed in recent years. An SMT system generates candidates with scores for all candidates and selects the sentence with the highest score as the correction result. However, the 1-best result of an SMT system is not always the best result. Thus, we propose a reranking approach for grammatical error correction. The reranking approach is used to re-score N-best results of the SMT and reorder the results. Our experiments show that our reranking system using parts of speech and syntactic features improves performance and achieves state-of-theart quality, with an F0.5 score of 40.0.

Most work on extracting parallel text from comparable corpora depends on linguistic resources such as seed parallel documents or translation dictionaries. This paper presents a simple baseline approach for bootstrapping a parallel collection. It starts by observing documents published on similar dates and the cooccurrence of a small number of identical tokens across languages. It then uses fast, online inference for a latent variable model to represent multilingual documents in a shared topic space where it can do efficient nearestneighbor search. Starting from the Gigaword collections in English and Spanish, we train a translation system that outperforms one trained on the WMT’11 parallel training set.

The Lexical Substitution task involves selecting and ranking lexical paraphrases for a target word in a given sentential context. We present PIC, a simple measure for estimating the appropriateness of substitutes in a given context. PIC outperforms another simple, comparable model proposed in recent work, especially when selecting substitutes from the entire vocabulary. Analysis shows that PIC improves over baselines by incorporating frequency biases into predictions.

Part-of-speech (POS) taggers trained on newswire perform much worse on domains such as subtitles, lyrics, or tweets. In addition, these domains are also heterogeneous, e.g., with respect to registers and dialects. In this paper, we consider the problem of learning a POS tagger for subtitles, lyrics, and tweets associated with African-American Vernacular English (AAVE). We learn from a mixture of randomly sampled and manually annotated Twitter data and unlabeled data, which we automatically and partially label using mined tag dictionaries. Our POS tagger obtains a tagging accuracy of 89% on subtitles, 85% on lyrics, and 83% on tweets, with up to 55% error reductions over a state-of-the-art newswire POS tagger, and 15-25% error reductions over a state-of-the-art Twitter POS tagger.

Previous work on Automatic Paraphrase Iden-tification (PI) is mainly based on modeling text similarity between two sentences. In contrast, we study methods for automatically detecting whether a text fragment only appearing in a sentence of the evaluated sentence pair is important or ancillary information with respect to the paraphrase identification task. Engineering features for this new task is rather difficult, thus, we approach the problem by representing text with syntactic structures and applying tree kernels on them. The results show that the accuracy of our automatic An-cillary Text Classifier (ATC) is promising, i.e., 68.6%, and its output can be used to improve the state of the art in PI.

In this paper, we explore sentiment composition in phrases that have at least one positive and at least one negative word—phrases like happy accident and best winter break. We compiled a dataset of such opposing polarity phrases and manually annotated them with real-valued scores of sentiment association. Using this dataset, we analyze the linguistic patterns present in opposing polarity phrases. Finally, we apply several unsupervised and supervised techniques of sentiment composition to determine their efficacy on this dataset. Our best system, which incorporates information from the phrase’s constituents, their parts of speech, their sentiment association scores, and their embedding vectors, obtains an accuracy of over 80% on the opposing polarity phrases.

We explore the consequences of representing token segmentations as hierarchical structures (trees) for the task of Multiword Expression (MWE) recognition, in isolation or in combination with dependency parsing. We propose a novel representation of token segmentation as trees on tokens, resembling dependency trees. Given this new representation, we present and evaluate two different architectures to combine MWE recognition and dependency parsing in the easy-first framework: a pipeline and a joint system, both taking advantage of lexical and syntactic dimensions. We experimentally validate that MWE recognition significantly helps syntactic parsing.

This work addresses the task of dependency labeling—assigning labels to an (unlabeled) dependency tree. We employ and extend a feature representation learning approach, optimizing it for both high speed and accuracy. We apply our labeling model on top of state-of-the-art parsers and evaluate its performance on standard benchmarks including the CoNLL-2009 and the English PTB datasets. Our model processes over 1,700 English sentences per second, which is 30 times faster than the sparse-feature method. It improves labeling accuracy over the outputs of top parsers, achieving the best LAS on 5 out of 7 datasets 1 .

Poorly translated text is often disfluent and difficult to read. In contrast, well-formed translations require less time to process. In this paper, we model the differences in reading patterns of Machine Translation (MT) evaluators using novel features extracted from their gaze data, and we learn to predict the quality scores given by those evaluators. We test our predictions in a pairwise ranking scenario, measuring Kendall’s tau correlation with the judgments. We show that our features provide information beyond fluency, and can be combined with BLEU for better predictions. Furthermore, our results show that reading patterns can be used to build semi -automatic metrics that anticipate the scores given by the evaluators.

This paper presents an study of the use of interlocking phrases in phrase-based statistical machine translation. We examine the effect on translation quality when the translation units used in the translation hypotheses are allowed to overlap on the source side, on the target side and on both sides. A large-scale evaluation on 380 language pairs was conducted. Our results show that overall the use of overlapping phrases improved translation quality by 0.3 BLEU points on average. Further analysis revealed that language pairs requiring a larger amount of re-ordering benefited the most from our approach. When the evaluation was restricted to such pairs, the average improvement increased to up to 0.75 BLEU points with over 97% of the pairs improving. Our approach requires only a simple modification to the decoding algorithm and we believe it should be generally applicable to improve the performance of phrase-based decoders.

We present a fast, simple, and high-accuracy short answer grading system. Given a short-answer question and its correct answer, key measures of the correctness of a student response can be derived from its semantic similarity with the correct answer. Our supervised model (1) utilizes recent advances in the identification of short-text similarity, and (2) augments text similarity features with key grading-specific constructs. We present experimental results where our model demonstrates top performance on multiple benchmarks.

Automated geolocation of social media messages can benefit a variety of downstream applications. However, these geolocation systems are typically evaluated without attention to how changes in time impact geolocation. Since different people, in different locations write messages at different times, these factors can significantly vary the performance of a ge-olocation system over time. We demonstrate cyclical temporal effects on geolocation accuracy in Twitter, as well as rapid drops as test data moves beyond the time period of training data. We show that temporal drift can effectively be countered with even modest online model updates.