ACL 2016 论文列表

We describe an efficient neural network method to automatically learn sentiment lexicons without relying on any manual resources. The method takes inspiration from the NRC method, which gives the best results in SemEval13 by leveraging emoticons in large tweets, using the PMI between words and tweet sentiments to define the sentiment attributes of words. We show that better lexicons can be learned by using them to predict the tweet sentiment labels. By using a very simple neural network, our method is fast and can take advantage of the same data volume as the NRC method. Experiments show that our lexicons give significantly better accuracies on multiple languages compared to the current best methods.

As a first step towards agents learning to communicate about their visual environment, we propose a system that, given visual representations of a referent (cat) and a context (sofa), identifies their discriminative attributes, i.e., properties that distinguish them (has_tail). Moreover, despite the lack of direct supervision at the attribute level, the model learns to assign plausible attributes to objects (sofa-has_cushion). Finally, we present a preliminary experiment confirming the referential success of the predicted discriminative attributes.

Relation classification is an important semantic processing task in the field of natural language processing (NLP). State-ofthe-art systems still rely on lexical resources such as WordNet or NLP systems like dependency parser and named entity recognizers (NER) to get high-level features. Another challenge is that important information can appear at any position in the sentence. To tackle these problems, we propose Attention-Based Bidirectional Long Short-Term Memory Networks(AttBLSTM) to capture the most important semantic information in a sentence. The experimental results on the SemEval-2010 relation classification task show that our method outperforms most of the existing methods, with only word vectors.

We demonstrate the value of collecting semantic parse labels for knowledge base question answering. In particular, (1) unlike previous studies on small-scale datasets, we show that learning from labeled semantic parses significantly improves overall performance, resulting in absolute 5 point gain compared to learning from answers, (2) we show that with an appropriate user interface, one can obtain semantic parses with high accuracy and at a cost comparable or lower than obtaining just answers, and (3) we have created and shared the largest semantic-parse labeled dataset to date in order to advance research in question answering.

,

This work is supported by ERC Consolidator Grant LEXICAL (648909) and KU Leuven Grant PDMK/14/117. SC is supported by ERC Starting Grant DisCoTex (306920).

Here we seek to automatically identify Hungarian patients suffering from mild cognitive impairment (MCI) based on linguistic features collected from their speech transcripts. Our system uses machine learning techniques and is based on several linguistic features like characteristics of spontaneous speech as well as features exploiting morphological and syntactic parsing. Our results suggest that it is primarily morphological and speech-based features that help distinguish MCI patients from healthy controls.

This work explores the use of unsupervised morph segmentation along with statistical language models for the task of vocabulary expansion. Unsupervised vocabulary expansion has large potential for improving vocabulary coverage and performance in different natural language processing tasks, especially in less-resourced settings on morphologically rich languages. We propose a combination of unsupervised morph segmentation and statistical language models and evaluate on languages from the Babel corpus. The method is shown to perform well for all the evaluated languages when compared to the previous work on the task.

tion is problematic for distortion-based alignment models. Alignments to NULL demand a special treatment as they would otherwise induce very long jumps that one does not usually observe in distortion-based alignment models. Examples of this can be found in Vogel et al. (1996), who drop the NULL word entirely and thus force all source words to align lexically, and Och and Ney (2003), who choose a fixed NULL probability. In the present work, we introduce a family of IBM-style alignment models that can express dependencies between translated and untranslated source words. The models do not use NULL words and instead allow untranslatable source words to be generated from translated words in their context. This is achieved by modelling source word collocations. From a technical point of view the model can be seen as a mixture of an alignment and a language model. In word alignment certain source words are only needed for fluency reasons and do not have a translation on the target side. Most word alignment models assume a target NULL word from which they generate these untranslatable source words. Hypothesising a target NULL word is not without problems, however. For example, because this NULL word has a position, it interferes with the distribution over alignment jumps. We present a word alignment model that accounts for untranslatable source words by generating them from preceding source words. It thereby removes the need for a target NULL word and only models alignments between word pairs that are actually observed in the data. Translation experiments on English paired with Czech, German, French and Japanese show that the model outperforms its traditional IBM counterparts in terms of BLEU score. 1 2 IBM models 1 and 2 Here, we quickly review the IBM alignment models 1 and 2 (Brown et al., 1993). We assume a random variable E over the English (target) vocabulary2 , a variable F over the French (source) vocabulary and a variable A over alignment links3 . The IBM models assign probabilities to alignment configurations and source sentences given the target side. Under the assumption that all source words are conditionally independent given the alignment links, these probabilities factorise as

We present an open web platform for developing, compiling, and running rulebased speech to sign language translation applications. Speech recognition is performed using the Nuance Recognizer 10.2 toolkit, and signed output, including both manual and non-manual components, is rendered using the JASigning avatar system. The platform is designed to make the component technologies readily accessible to sign language experts who are not necessarily computer scientists. Translation grammars are written in a version of Synchronous Context-Free Grammar adapted to the peculiarities of sign language. All processing is carried out on a remote server, with content uploaded and accessed through a web interface. Initial experiences show that simple translation grammars can be implemented on a time-scale of a few hours to a few days and produce signed output readily comprehensible to Deaf informants. Overall, the platform drastically lowers the barrier to entry for researchers interested in building applications that generate high-quality signed language.

We introduce a new multilingual resource containing judgments about nominal compound compositionality in English, French and Portuguese. It covers 3 × 180 noun-noun and adjective-noun compounds for which we provide numerical compositionality scores for the head word, for the modifier and for the compound as a whole, along with possible paraphrases. This resource was constructed by native speakers via crowdsourcing. It can serve as basis for evaluating tasks such as lexical substitution and compositionality prediction.

Named entity recognition, and other information extraction tasks, frequently use linguistic features such as part of speech tags or chunkings. For languages where word boundaries are not readily identified in text, word segmentation is a key first step to generating features for an NER system. While using word boundary tags as features are helpful, the signals that aid in identifying these boundaries may provide richer information for an NER system. New state-of-the-art word segmentation systems use neural models to learn representations for predicting word boundaries. We show that these same representations, jointly trained with an NER system, yield significant improvements in NER for Chinese social media. In our experiments, jointly training NER and word segmentation with an LSTM-CRF model yields nearly 5% absolute improvement over previously published results.

We release the Simple Paraphrase Database, a subset of of the Paraphrase Database (PPDB) adapted for the task of text simplification. We train a supervised model to associate simplification scores with each phrase pair, producing rankings competitive with state-of-theart lexical simplification models. Our new simplification database contains 4.5 million paraphrase rules, making it the largest available resource for lexical simplification.

In this paper we improve over the hierarchical Pitman-Yor processes language model in a cross-domain setting by adding skipgrams as features. We find that adding skipgram features reduces the perplexity. This reduction is substantial when models are trained on a generic corpus and tested on domain-specific corpora. We also find that within-domain testing and cross-domain testing require different backoff strategies. We observe a 30-40% reduction in perplexity in a cross-domain language modelling task, and up to 6% reduction in a within-domain experiment, for both English and Flemish-Dutch.

In this paper, we propose the TBCNN-pair model to recognize entailment and contradiction between two sentences. In our model, a tree-based convolutional neural network (TBCNN) captures sentence-level semantics; then heuristic matching layers like concatenation, element-wise product/difference combine the information in individual sentences. Experimental results show that our model outperforms existing sentence encoding-based approaches by a large margin.

In order to capture rich language phenomena, neural machine translation models have to use a large vocabulary size, which requires high computing time and large memory usage. In this paper, we alleviate this issue by introducing a sentence-level or batch-level vocabulary, which is only a very small sub-set of the full output vocabulary. For each sentence or batch, we only predict the target words in its sentence-level or batch-level vocabulary. Thus, we reduce both the computing time and the memory usage. Our method simply takes into account the translation options of each word or phrase in the source sentence, and picks a very small target vocabulary for each sentence based on a word-to-word translation model or a bilingual phrase library learned from a traditional machine translation model. Experimental results on the large-scale English-to-French task show that our method achieves better translation performance by 1 BLEU point over the large vocabulary neural machine translation system of Jean et al. (2015).

In unsupervised semantic role labeling, identifying the role of an argument is usually informed by its dependency relation with the predicate. In this work, we pro-pose a neural model to learn argument embeddings from the context by explicitly incorporating dependency relations as multiplicative factors, which bias argu-ment embeddings according to their dependency roles. Our model outperforms existing state-of-the-art embeddings in un-supervised semantic role induction on the CoNLL 2008 dataset and the SimLex999 word similarity task. Qualitative results demonstrate our model can effectively bias argument embeddings based on their dependency role.

Recent work in learning vector-space embeddings for multi-relational data has focused on combining relational information derived from knowledge bases with distributional information derived from large text corpora. We propose a simple approach that leverages the descriptions of entities or phrases available in lexical resources, in conjunction with distributional semantics, in order to derive a better initialization for training relational models. Applying this initialization to the TransE model results in significant new state-of-the-art performances on the WordNet dataset, decreasing the mean rank from the previous best of 212 to 51. It also results in faster convergence of the entity representations. We find that there is a trade-off between improving the mean rank and the hits@10 with this approach. This illustrates that much remains to be understood regarding performance improvements in relational models.

Over the past decade, e-Commerce has rapidly grown enabling customers to purchase products with the click of a button. But to be able to do so, one has to understand the semantics of a user query and identify that in digital lifestyle tv , digital lifestyle is a brand and tv is a product. In this paper, we develop a series of structured prediction algorithms for semantic tagging of shopping queries with the product , brand , model and product family types. We model wide variety of features and show an alternative way to capture knowledge base information using embed-dings. We conduct an extensive study over 37 , 000 manually annotated queries and report performance of 90 . 92 F 1 independent of the query length.