NAACL 2015 Papers

We present our work on leveraging multilingual parallel corpora of small sizes for Statistical Machine Translation between Japanese and Hindi using multiple pivot languages. In our setting, the source and target part of the corpus remains the same, but we show that using several different pivot to extract phrase pairs from these source and target parts lead to large BLEU improvements. We focus on a variety of ways to exploit phrase tables generated using multiple pivots to support a direct source-target phrase table. Our main method uses the Multiple Decoding Paths (MDP) feature of Moses, which we empirically verify as the best compared to the other methods we used. We compare and contrast our various results to show that one can overcome the limitations of small corpora by using as many pivot languages as possible in a multilingual setting. Most importantly, we show that such pivoting aids in learning of additional phrase pairs which are not learned when the direct sourcetarget corpus is small. We obtained improvements of up to 3 BLEU points using multiple pivots for Japanese to Hindi translation compared to when only one pivot is used. To the best of our knowledge, this work is also the first of its kind to attempt the simultaneous utilization of 7 pivot languages at decoding time.

Evaluation of segment-level machine translation metrics is currently hampered by: (1) low inter-annotator agreement levels in human assessments; (2) lack of an effective mechanism for evaluation of translations of equal quality; and (3) lack of methods of significance testing improvements over a baseline. In this paper, we provide solutions to each of these challenges and outline a new human evaluation methodology aimed specifically at assessment of segment-level metrics. We replicate the human evaluation component of WMT-13 and reveal that the current state-of-the-art performance of segment-level metrics is better than previously believed. Three segment-level metrics — METEOR, NLEPOR and SENTBLEUMOSES — are found to correlate with human assessment at a level not significantly outperformed by any other metric in both the individual language pair assessment for Spanish-toEnglish and the aggregated set of 9 language pairs.

Current approaches to cross-lingual information retrieval (CLIR) rely on standard retrieval models into which query translations by statistical machine translation (SMT) are integrated at varying degree. In this paper, we present an attempt to turn this situation on its head: Instead of the retrieval aspect, we emphasize the translation component in CLIR. We perform search by using an SMT decoder in forced decoding mode to produce a bag-ofwords representation of the target documents to be ranked. The SMT model is extended by retrieval-specific features that are optimized jointly with standard translation features for a ranking objective. We find significant gains over the state-of-the-art in a large-scale evaluation on cross-lingual search in the domains patents and Wikipedia.

Extensive lexical knowledge is necessary for temporal analysis and planning tasks. We address in this paper a lexical setting that allows for the straightforward incorporation of rich features and structural constraints. We explore a lexical event ordering task, namely determining the likely temporal order of events based solely on the identity of their predicates and arguments. We propose an “edgefactored” model for the task that decomposes over the edges of the event graph. We learn it using the structured perceptron. As lexical tasks require large amounts of text, we do not attempt manual annotation and instead use the textual order of events in a domain where this order is aligned with their temporal order, namely cooking recipes.

This paper introduces a tensor-based approach to semantic role labeling (SRL). The motivation behind the approach is to automatically induce a compact feature representation for words and their relations, tailoring them to the task. In this sense, our dimensionality reduction method provides a clear alternative to the traditional feature engineering approach used in SRL. To capture meaningful interactions between the argument, predicate, their syntactic path and the corresponding role label, we compress each feature representation first to a lower dimensional space prior to assessing their interactions. This corresponds to using an overall cross-product feature representation and maintaining associated parameters as a four-way low-rank tensor. The tensor parameters are optimized for the SRL performance using standard online algorithms. Our tensor-based approach rivals the best performing system on the CoNLL-2009 shared task. In addition, we demonstrate that adding the representation tensor to a competitive tensorfree model yields 2% absolute increase in Fscore. 1

This paper describes IDEST, a new method for learning paraphrases of event patterns. It is based on a new neural network architecture that only relies on the weak supervision signal that comes from the news published on the same day and mention the same real-world entities. It can generalize across extractions from different dates to produce a robust paraphrase model for event patterns that can also capture meaningful representations for rare patterns. We compare it with two state-of-the-art systems and show that it can attain comparable quality when trained on a small dataset. Its generalization capabilities also allow it to leverage much more data, leading to substantial quality improvements.

Most successful Entity Linking (EL) methods aim to link mentions to their referent entities in a structured Knowledge Base (KB) by comparing their respective contexts, often using similarity measures. While the KB structure is given, current methods have suffered from impoverished information representations on the mention side. In this paper, we demonstrate the effectiveness of Abstract Meaning Representation (AMR) (Banarescu et al., 2013) to select high quality sets of entity “collaborators” to feed a simple similarity measure (Jaccard) to link entity mentions. Experimental results show that AMR captures contextual properties discriminative enough to make linking decisions, without the need for EL training data, and that system with AMR parsing output outperforms hand labeled traditional semantic roles as context representation for EL. Finally, we show promising preliminary results for using AMR to select sets of “coherent” entity mentions for collective entity linking 1 .

Matrix factorization approaches to relation extraction provide several attractive features: they support distant supervision, handle open schemas, and leverage unlabeled data. Unfortunately, these methods share a shortcoming with all other distantly supervised approaches: they cannot learn to extract target relations without existing data in the knowledge base, and likewise, these models are inaccurate for relations with sparse data. Rule-based extractors, on the other hand, can be easily extended to novel relations and improved for existing but inaccurate relations, through first-order formulae that capture auxiliary domain knowledge. However, usually a large set of such formulae is necessary to achieve generalization. In this paper, we introduce a paradigm for learning low-dimensional embeddings of entity-pairs and relations that combine the advantages of matrix factorization with first-order logic domain knowledge. We introduce simple approaches for estimating such embeddings, as well as a novel training algorithm to jointly optimize over factual and first-order logic information. Our results show that this method is able to learn accurate extractors with little or no distant supervision alignments, while at the same time generalizing to textual patterns that do not appear in the formulae.

Coreference is a core nlp problem. However, newswire data, the primary source of existing coreference data, lack the richness necessary to truly solve coreference. We present a new domain with denser references—quiz bowl questions—that is challenging and enjoyable to humans, and we use the quiz bowl community to develop a new coreference dataset, together with an annotation framework that can tag any text data with coreferences and named entities. We also successfully integrate active learning into this annotation pipeline to collect documents maximally useful to coreference models. State-of-the-art coreference systems underperform a simple classifier on our new dataset, motivating non-newswire data for future coreference research.

Recent work has successfully leveraged the semantic information extracted from lexical knowledge bases such as WordNet and FrameNet to improve English event coreference resolvers. The lack of comparable resources in other languages, however, has made the design of high-performance non-English event coreference resolvers, particularly those employing unsupervised models, very difficult. We propose a generative model for the under-studied task of Chinese event coreference resolution that rivals its supervised counterparts in performance when evaluated on the ACE 2005 corpus.

Previous work on text coherence was primarily based on matching multiple mentions of the same entity in di erent parts of the text; therefore, it misses the contribution from semantically related but not necessarily coreferential entities (e.g., Gates and Microsoft). In this paper, we capture such semantic relatedness by leveraging world knowledge (e.g., Gates is the person who created Microsoft), and use two existing evaluation frameworks. First, in the unsupervised framework, we introduce semantic relatedness as an enrichment to the original graph-based model of Guinaudeau and Strube (2013). In addition, we incorporate semantic relatedness as additional features into the popular entity-based model of Barzilay and Lapata (2008). Across both frameworks, our enriched model with semantic relatedness outperforms the original methods, especially on short documents.

We present a novel abstractive summarization framework that draws on the recent development of a treebank for the Abstract Meaning Representation (AMR). In this framework, the source text is parsed to a set of AMR graphs, the graphs are transformed into a summary graph, and then text is generated from the summary graph. We focus on the graph-tograph transformation that reduces the source semantic graph into a summary graph, making use of an existing AMR parser and assuming the eventual availability of an AMR-totext generator. The framework is data-driven, trainable, and not specifically designed for a particular domain. Experiments on goldstandard AMR annotations and system parses show promising results. Code is available at: https://github.com/summarization

In this paper we study the task of movie script summarization, which we argue could enhance script browsing, give readers a rough idea of the script’s plotline, and speed up reading time. We formalize the process of generating a shorter version of a screenplay as the task of finding an optimal chain of scenes. We develop a graph-based model that selects a chain by jointly optimizing its logical progression, diversity, and importance. Human evaluation based on a question-answering task shows that our model produces summaries which are more informative compared to competitive baselines.

Existing timeline generation systems for complex events consider only information from traditional media, ignoring the rich social context provided by user-generated content that reveals representative public interests or insightful opinions. We instead aim to generate socially-informed timelines that contain both news article summaries and selected user comments. We present an optimization framework designed to balance topical cohesion between the article and comment summaries along with their informativeness and coverage of the event. Automatic evaluations on real-world datasets that cover four complex events show that our system produces more informative timelines than state-of-the-art systems. In human evaluation, the associated comment summaries are furthermore rated more insightful than editor's picks and comments ranked highly by users.

A major opportunity for NLP to have a realworld impact is in helping educators score student writing, particularly content-based writing (i.e., the task of automated short answer scoring). A major challenge in this enterprise is that scored responses to a particular question (i.e., labeled data) are valuable for modeling but limited in quantity. Additional information from the scoring guidelines for humans, such as exemplars for each score level and descriptions of key concepts, can also be used. Here, we explore methods for integrating scoring guidelines and labeled responses, and we find that stacked generalization (Wolpert, 1992) improves performance, especially for small training sets.

Citation sentences (citances) to a reference article have been extensively studied for summarization tasks. However, citances might not accurately represent the content of the cited article, as they often fail to capture the context of the reported findings and can be affected by epistemic value drift. Following the intuition behind the TAC (Text Analysis Conference) 2014 Biomedical Summarization track, we propose a system that identifies text spans in the reference article that are related to a given citance. We refer to this problem as citance-reference spans matching. We approach the problem as a retrieval task; in this paper, we detail a comparison of different citance reformulation methods and their combinations. While our results show improvement over the baseline (up to 25.9%), their absolute magnitude implies that there is ample room for future improvement.

Transcribing documents from the printing press era, a challenge in its own right, is more complicated when documents interleave multiple languages—a common feature of 16th century texts. Additionally, many of these documents precede consistent orthographic conventions, making the task even harder. We extend the state-of-the-art historical OCR model of Berg-Kirkpatrick et al. (2013) to handle word-level code-switching between multiple languages. Further, we enable our system to handle spelling variability, including now-obsolete shorthand systems used by printers. Our results show average relative character error reductions of 14% across a variety of historical texts.

We present the first dynamic programming (DP) algorithm for shift-reduce constituency parsing, which extends the DP idea of Huang and Sagae (2010) to context-free grammars. To alleviate the propagation of errors from part-of-speech tagging, we also extend the parser to take a tag lattice instead of a fixed tag sequence. Experiments on both English and Chinese treebanks show that our DP parser significantly improves parsing quality over non-DP baselines, and achieves the best accuracies among empirical linear-time parsers.

Supervised morphological paradigm learning by identifying and aligning the longest common subsequence found in inflection tables has recently been proposed as a simple yet competitive way to induce morphological patterns. We combine this non-probabilistic strategy of inflection table generalization with a discriminative classifier to permit the reconstruction of complete inflection tables of unseen words. Our system learns morphological paradigms from labeled examples of inflection patterns (inflection tables) and then produces inflection tables from unseen lemmas or base forms. We evaluate the approach on datasets covering 11 different languages and show that this approach results in consistently higher accuracies vis-` other methods on the same task, thus indicating that the general method is a viable approach to quickly creating highaccuracy morphological resources.