NAACL 2016 论文列表

Understanding how a fictional relationship between two characters changes over time (e.g., from best friends to sworn enemies) is a key challenge in digital humanities scholarship. We present a novel unsupervised neural network for this task that incorporates dictionary learning to generate interpretable, accurate relationship trajectories. While previous work on characterizing literary relationships relies on plot summaries annotated with predefined labels, our model jointly learns a set of global relationship descriptors as well as a trajectory over these descriptors for each relationship in a dataset of raw text from novels. We find that our model learns descriptors of events (e.g., marriage or murder) as well as interpersonal states (love, sadness). Our model outperforms topic model baselines on two crowdsourced tasks, and we also find interesting correlations to annotations in an existing dataset.

We show how eye-tracking corpora can be used to improve sentence compression models, presenting a novel multi-task learning algorithm based on multi-layer LSTMs. We obtain performance competitive with or better than state-of-the-art approaches.

We introduce a novel, simple convolution neural network (CNN) architecture - multi-group norm constraint CNN (MGNC-CNN) that capitalizes on multiple sets of word embeddings for sentence classification. MGNC-CNN extracts features from input embedding sets independently and then joins these at the penultimate layer in the network to form a final feature vector. We then adopt a group regularization strategy that differentially penalizes weights associated with the subcomponents generated from the respective embedding sets. This model is much simpler than comparable alternative architectures and requires substantially less training time. Furthermore, it is flexible in that it does not require input word embeddings to be of the same dimensionality. We show that MGNC-CNN consistently outperforms baseline models.

The goal of sentence and document modeling is to accurately represent the meaning of sentences and documents for various Natural Language Processing tasks. In this work, we present Dependency Sensitive Convolutional Neural Networks (DSCNN) as a generalpurpose classification system for both sentences and documents. DSCNN hierarchically builds textual representations by processing pretrained word embeddings via Long ShortTerm Memory networks and subsequently extracting features with convolution operators. Compared with existing recursive neural models with tree structures, DSCNN does not rely on parsers and expensive phrase labeling, and thus is not restricted to sentencelevel tasks. Moreover, unlike other CNNbased models that analyze sentences locally by sliding windows, our system captures both the dependency information within each sentence and relationships across sentences in the same document. Experiment results demonstrate that our approach is achieving state-ofthe-art performance on several tasks, including sentiment analysis, question type classification, and subjectivity classification.

Traditional convolutional neural network ( CNN ) based query classification uses linear feature mapping in its convolution operation. The recurrent neural network ( RNN ), differs from a CNN in representing word sequence with their ordering information kept explicitly. We propose using a deep long-short-term-memory ( DLSTM ) based feature mapping to learn feature representation for CNN . The DLSTM , which is a stack of LSTM units, has different order of feature representations at different depth of LSTM unit. The bottom LSTM unit equipped with input and output gates, extracts the first order feature representation from current word. To extract higher order nonlinear feature representation, the LSTM unit at higher position gets input from two parts. First part is the lower LSTM unit’s memory cell from previous word. Second part is the lower LSTM unit’s hidden output from current word. In this way, the DLSTM captures the nonlinear nonconsecutive interaction within n -grams. Using an architecture that combines a stack of the DLSTM layers with a tradition CNN layer, we have observed new state-of-the-art query classification accuracy on benchmark data sets for query classification.

We compare different word embeddings from a standard window based skipgram model, a skipgram model trained using dependency context features and a novel skipgram variant that utilizes additional information from dependency graphs. We explore the effectiveness of the different types of word embeddings for word similarity and sentence classification tasks. We consider three common sentence classification tasks: question type classification on the TREC dataset, binary sentiment classification on Stanford’s Sentiment Treebank and semantic relation classification on the SemEval 2010 dataset. For each task we use three different classification methods: a Support Vector Machine, a Convolutional Neural Network and a Long Short Term Memory Network. Our experiments show that dependency based embeddings outperform standard window based embeddings in most of the settings, while using dependency context embeddings as additional features improves performance in all tasks regardless of the classification method. Ourembeddings and code are available at https:

We propose a hierarchical attention network for document classification. Our model has two distinctive characteristics: (i) it has a hierarchical structure that mirrors the hierarchical structure of documents; (ii) it has two levels of attention mechanisms applied at the wordand sentence-level, enabling it to attend differentially to more and less important content when constructing the document representation. Experiments conducted on six large scale text classification tasks demonstrate that the proposed architecture outperform previous methods by a substantial margin. Visualization of the attention layers illustrates that the model selects qualitatively informative words and sentences.

Gold annotations for supervised implicit semantic role labeling are extremely sparse and costly. As a lightweight alternative, this paper describes an approach based on unsupervised parsing which can do without iSRL-specific training data: We induce prototypical roles from large amounts of explicit SRL annotations paired with their distributed word representations. An evaluation shows competitive performance with supervised methods on the SemEval 2010 data, and our method can easily be applied to predicates (or languages) for which no training annotations are available.

Automatically generated databases of English paraphrases have the drawback that they return a single list of paraphrases for an input word or phrase. This means that all senses of polysemous words are grouped together, unlike WordNet which partitions different senses into separate synsets. We present a new method for clustering paraphrases by word sense, and apply it to the Paraphrase Database (PPDB). We investigate the performance of hierarchical and spectral clustering algorithms, and systematically explore different ways of defining the similarity matrix that they use as input. Our method produces sense clusters that are qualitatively and quantitatively good, and that represent a substantial improvement to the PPDB resource.

We introduce a latent activity model for workplace emails, positing that communication at work is purposeful and organized by activities. We pose the problem as probabilistic inference in graphical models that jointly capture the interplay between latent activities and the email contexts they govern, such as the recipients, subject and body. The model parameters are learned using maximum likelihood estimation with an expectation maximization algorithm. We present three variants of the model that incorporate the recipients, co-occurrence of the recipients, and email body and subject. We demonstrate the model’s effectiveness in an email recipient recommendation task and show that it outperforms a state-of-the-art generative model. Additionally, we show that the activity model can be used to identify email senders who engage in similar activities, resulting in further improvements in recipient recommendation.

Natural language inference (NLI) is a fundamentally important task in natural language processing that has many applications. The recently released Stanford Natural Language Inference (SNLI) corpus has made it possible to develop and evaluate learning-centered methods such as deep neural networks for natural language inference (NLI). In this paper, we propose a special long short-term memory (LSTM) architecture for NLI. Our model builds on top of a recently proposed neural attention model for NLI but is based on a significantly different idea. Instead of deriving sentence embeddings for the premise and the hypothesis to be used for classification, our solution uses a match-LSTM to perform word-by-word matching of the hypothesis with the premise. This LSTM is able to place more emphasis on important word-level matching results. In particular, we observe that this LSTM remembers important mismatches that are critical for predicting the contradiction or the neutral relationship label. On the SNLI corpus, our model achieves an accuracy of 86.1%, outperforming the state of the art.

This paper presents a methodology to extract positive interpretations from negated statements.

While there is increasing interest in automatically recognizing the argumentative structure of a text, recognizing the argumentative purpose of revisions to such texts has been less explored. Furthermore, existing revision clas-sification approaches typically ignore contextual information. We propose two approaches for utilizing contextual information when predicting argumentative revision purposes: developing contextual features for use in the classification paradigm of prior work, and transforming the classification problem to a sequence labeling task. Experimental results using two corpora of student essays demonstrate the utility of contextual information for predicting argumentative revision purposes.

Coherence is established by semantic connections between sentences of a text which can be modeled by lexical relations. In this paper, we introduce the lexical coherence graph (LCG), a new graph-based model to represent lexical relations among sentences. The frequency of subgraphs (coherence patterns) of this graph captures the connectivity style of sentence nodes in this graph. The coherence of a text is encoded by a vector of these frequencies. We evaluate the LCG model on the readability ranking task. The results of the experiments show that the LCG model obtains higher accuracy than state-of-the-art coherence models. Using larger subgraphs yields higher accuracy, because they capture more structural information. However, larger subgraphs can be sparse. We adapt Kneser-Ney smoothing to smooth subgraphs’ frequencies. Smoothing improves performance.

Persuasive communication is the process of shaping, reinforcing and changing others' responses. In political debates, speakers express their views towards the debated topics by choosing both the content of their discourse and the argumentation process. In this work we study the use of semantic frames for modelling argumentation in speakers' discourse. We investigate the impact of a speaker's argumentation style and their effect in influencing an audience in supporting their candidature. We model the influence index of each candidate based on their relative standings in the polls released prior to the debate and present a system which ranks speakers in terms of their relative influence using a combination of content and persuasive argumentation features. Our results show that although content alone is predictive of a speaker's influence rank, persuasive argumentation also affects such indices.

Argumentation mining is considered as a key technology for future search engines and automated decision making. In such applications, argumentative text segments have to be mined from large and diverse document collections. However, most existing argumentation mining approaches tackle the classification of argumentativeness only for a few manually annotated documents from narrow domains and registers. This limits their practical applicability. We hence propose a distant supervision approach that acquires argumen-tative text segments automatically from online debate portals. Experiments across domains and registers show that training on such a corpus improves the effectiveness and robustness of mining argumentative text. We freely provide the underlying corpus for research.

Understanding the argumentative structure of a persuasive essay involves addressing two challenging tasks: identifying the components of the essay’s argument and identifying the relations that occur between them. We examine the under-investigated task of end-to-end argument mining in persuasive student essays, where we (1) present the first results on end-to-end argument mining in student essays using a pipeline approach; (2) address error propagation inherent in the pipeline approach by performing joint inference over the outputs of the tasks in an Integer Linear Programming (ILP) framework; and (3) propose a novel objective function that enables F-score to be maximized directly by an ILP solver. We evaluate our joint-inference approach with our novel objective function on a publicly-available corpus of 90 essays, where it yields an 18.5% relative error reduction in F-score over the pipeline system.

Jauhar et al. (2015) recently proposed to learn sense-specific word representations by “retrofitting” standard distributional word representations to an existing ontology. We observe that this approach does not require an ontology

Unsupervised methods for learning distributed representations of words are ubiquitous in today's NLP research, but far less is known about the best ways to learn distributed phrase or sentence representations from unlabelled data. This paper is a systematic comparison of models that learn such representations. We find that the optimal approach depends critically on the intended application. Deeper, more complex models are preferable for representations to be used in supervised systems, but shallow log-linear models work best for building representation spaces that can be decoded with simple spatial distance metrics. We also propose two new unsupervised representation-learning objectives designed to optimise the trade-off between training time, domain portability and performance.