EMNLP 2015 Papers

We study the problem of jointly embedding a knowledge base and a text corpus. The key issue is the alignment model making sure the vectors of entities, relations and words are in the same space. Wang et al. (2014a) rely on Wikipedia anchors, making the applicable scope quite limited. In this paper we propose a new alignment model based on text descriptions of entities, without dependency on anchors. We require the embedding vector of an entity not only to fit the structured constraints in KBs but also to be equal to the embedding vector computed from the text description. Extensive experiments show that, the proposed approach consistently performs comparably or even better than the method of Wang et al. (2014a), which is encouraging as we do not use any anchor information.

A wide range of applications, from social media to scientific literature analysis, involve graphs in which documents are connected by links. We introduce a topic model for link prediction based on the intuition that linked documents will tend to have similar topic distributions, integrating a max-margin learning criterion and lexical term weights in the loss function. We validate our approach on the tweets from 2,000 Sina Weibo users and evaluate our model’s reconstruction of the social network.

In the last several years, neural network models have significantly improved accuracy in a number of NLP tasks. However, one serious drawback that has impeded their adoption in production systems is the slow runtime speed of neural network models compared to alternate models, such as maximum entropy classifiers. In Devlin et al. (2014), the authors presented a simple technique for speeding up feed-forward embedding-based neural network models, where the dot product between each word embedding and part of the first hidden layer are pre-computed offline. However, this technique cannot be used for hidden layers beyond the first. In this paper, we explore a neural network architecture where the embedding layer feeds into multiple hidden layers that are placed “next to” one another so that each can be pre-computed independently. On a large scale language modeling task, this architecture achieves a 10x speedup at runtime and a significant reduction in perplexity when compared to a standard multilayer network.

Research on modeling time series text corpora has typically focused on predicting what text will come next, but less well studied is predicting when the next text event will occur. In this paper we address the latter case, framed as modeling continuous inter-arrival times under a logGaussian Cox process, a form of inhomogeneous Poisson process which captures the varying rate at which the tweets arrive over time. In an application to rumour modeling of tweets surrounding the 2014 Ferguson riots, we show how interarrival times between tweets can be accurately predicted, and that incorporating textual features further improves predictions.

When building spoken dialogue systems for a new domain, a major bottleneck is developing a spoken language understanding (SLU) module that handles the new domain’s terminology and semantic concepts. We propose a statistical SLU model that generalises to both previously unseen input words and previously unseen output classes by leveraging unlabelled data. After mapping the utterance into a vector space, the model exploits the structure of the output labels by mapping each label to a hyperplane that separates utterances with and without that label. Both these mappings are initialised with unsupervised word embeddings, so they can be computed even for words or concepts which were not in the SLU training data.

We investigate an extension of continuous online learning in recurrent neural network language models. The model keeps a separate vector representation of the current unit of text being processed and adaptively adjusts it after each prediction. The initial experiments give promising results, indicating that the method is able to increase language modelling accuracy, while also decreasing the parameters needed to store the model along with the computation required at each step.

In this paper we explore a POS tagging application of neural architectures that can infer word representations from the raw character stream. It relies on two modelling stages that are jointly learnt: a convolutional network that infers a word representation directly from the character stream, followed by a prediction stage. Models are evaluated on a POS and morphological tagging task for German. Experimental results show that the convolutional network can infer meaningful word representations, while for the prediction stage, a well designed and structured strategy allows the model to outperform stateof-the-art results, without any feature engineering.

This paper discusses the use of factorization techniques in distributional semantic models. We focus on a method for redistributing the weight of latent variables, which has previously been shown to improve the performance of distributional semantic models. However, this result has not been replicated and remains poorly understood. We refine the method, and provide additional theoretical justification, as well as empirical results that demonstrate the viability of the proposed approach.

IBM Model 1 is a classical alignment model. Of the first generation word-based SMT models, it was the only such model with a concave objective function. For concave optimization problems like IBM Model 1, we have guarantees on the convergence of optimization algorithms such as Expectation Maximization (EM). However, as was pointed out recently, the objective of IBM Model 1 is not strictly concave and there is quite a bit of alignment quality variance within the optimal solution set. In this work we detail a strictly concave version of IBM Model 1 whose EM algorithm is a simple modification of the original EM algorithm of Model 1 and does not require the tuning of a learning rate or the insertion of an l2 penalty. Moreover, by addressing Model 1’s shortcomings, we achieve AER and F-Measure improvements over the classical Model 1 by over 30%.

We investigate the role that geometric, textual and visual features play in the task of predicting a preposition that links two visual entities depicted in an image. The task is an important part of the subsequent process of generating image descriptions. We explore the prediction of prepositions for a pair of entities, both in the case when the labels of such entities are known and unknown. In all situations we found clear evidence that all three features contribute to the prediction task.

Integrating vision and language has long been a dream in work on artificial intelligence (AI). In the past two years, we have witnessed an explosion of work that brings together vision and language from images to videos and beyond. The available corpora have played a crucial role in advancing this area of research. In this paper, we propose a set of quality metrics for evaluating and analyzing the vision & language datasets and categorize them accordingly. Our analyses show that the most recent datasets have been using more complex language and more abstract concepts, however, there are different strengths and weaknesses in each.

In this paper we focus on a new problem of event coreference resolution across television news videos. Based on the observation that the contents from multiple data modalities are complementary, we develop a novel approach to jointly encode effective features from both closed captions and video key frames. Experiment results demonstrate that visual features provided 7.2% absolute F-score gain on stateof-the-art text based event extraction and coreference resolution.

A simile is a comparison between two essentially unlike things, such as “Jane swims like a dolphin”. Similes often express a positive or negative sentiment toward something, but recognizing the polarity of a simile can depend heavily on world knowledge. For example, “memory like an elephant” is positive, but “memory like a sieve” is negative. Our research explores methods to recognize the polarity of similes on Twitter. We train classifiers using lexical, semantic, and sentiment features, and experiment with both manually and automatically generated training data. Our approach yields good performance at identifying positive and negative similes, and substantially outperforms existing sentiment resources.

In this work, we build an entity/event-level sentiment analysis system, which is able to recognize and infer both explicit and implicit sentiments toward entities and events in the text. We design Probabilistic Soft Logic models that integrate explicit sentiments, inference rules, and +/-effect event information (events that positively or negatively affect entities). The experiments show that the method is able to greatly improve over baseline accuracies in recognizing entity/event-level sentiments.

Opinion summarization is the task of producing the summary of a text, such that the summary also preserves the sentiment of the text. Opinion Summarization is thus a trade-off between summarization and sentiment analysis. The demand of compression may drop sentiment bearing sentences, and the demand of sentiment detection may bring in redundant sentences. We harness the power of submodularity to strike a balance between two conflicting requirements. We investigate an incipient class of submodular functions for the problem, and a partial enumeration based greedy algorithm that has performance guarantee of 63%. Our functions generate summaries such that there is good correlation between document sentiment and summary sentiment along with good ROUGE score, which outperforms thestate-of-the-art algorithms.

Cross-Lingual Learning provides a mechanism to adapt NLP tools available for label rich languages to achieve similar tasks for label-scarce languages. An efficient cross-lingual tool significantly reduces the cost and effort required to manually annotate data. In this paper, we use the Recursive Autoencoder architecture to develop a Cross Lingual Sentiment Analysis (CLSA) tool using sentence aligned corpora between a pair of resource rich (English) and resource poor (Hindi) language. The system is based on the assumption that semantic similarity between different phrases also implies sentiment similarity in majority of sentences. The resulting system is then analyzed on a newly developed Movie Reviews Dataset in Hindi with labels given on a rating scale and compare performance of our system against existing systems. It is shown that our approach significantly outperforms state of the art systems for Sentiment Analysis, especially when labeled data is scarce.

This paper is concerned with the task of bilingual lexicon induction using imagebased features. By applying features from a convolutional neural network (CNN), we obtain state-of-the-art performance on a standard dataset, obtaining a 79% relative improvement over previous work which uses bags of visual words based on SIFT features. The CNN image-based approach is also compared with state-of-the-art linguistic approaches to bilingual lexicon induction, even outperforming these for one of three language pairs on another standard dataset. Furthermore, we shed new light on the type of visual similarity metric to use for genuine similarity versus relatedness tasks, and experiment with using multiple layers from the same network in an attempt to improve performance.

Social media such as Twitter have become an important method of communication, with potential opportunities for NLG to facilitate the generation of social media content. We focus on the generation of indicative tweets that contain a link to an external web page. While it is natural and tempting to view the linked web page as the source text from which the tweet is generated in an extractive summarization setting, it is unclear to what extent actual indicative tweets behave like extractive summaries. We collect a corpus of indicative tweets with their associated articles and investigate to what extent they can be derived from the articles using extractive methods. We also consider the impact of the formality and genre of the article. Our results demonstrate the limits of viewing indicative tweet generation as extractive summarization, and point to the need for the development of a methodology for tweet generation that is sensitive to genre-specific issues.

We provide an analysis of current evaluation methodologies applied to summarization metrics and identify the following areas of concern: (1) movement away from evaluation by correlation with human assessment; (2) omission of important components of human assessment from evaluations, in addition to large numbers of metric variants; (3) absence of methods of significance testing improvements over a baseline. We outline an evaluation methodology that overcomes all such challenges, providing the first method of significance testing suitable for evaluation of summarization metrics. Our evaluation reveals for the first time which metric variants significantly outperform others, optimal metric variants distinct from current recommended best variants, as well as machine translation metric BLEU to have performance on-par with ROUGE for the purpose of evaluation of summarization systems. We subsequently replicate a recent large-scale evaluation that relied on, what we now know to be, suboptimal ROUGE variants revealing distinct conclusions about the relative performance of state-of-the-art summarization systems.