Recent approaches to learning structured predictors often require approximate inference for tractability; yet its effects on the learned model are unclear. Meanwhile, most learning algorithms act as if computational cost was constant within the model class. This paper sheds some light on the first issue by establishing risk bounds for max-margin learning with LP relaxed inference and addresses the second issue by proposing a new paradigm that attempts to penalize “timeconsuming” hypotheses. Our analysis relies on a geometric characterization of the outer polyhedra associated with the LP relaxation. We then apply these techniques to the problem of dependency parsing, for which a concise LP formulation is provided that handles non-local output features. A significant improvement is shown over arc-factored models. 1.

Text summarization is one of the oldest problems in natural language processing. Popular approaches rely on extracting relevant sentences from the original documents. As a side effect, sentences that are too long but partly relevant are doomed to either not appear in the final summary, or prevent inclusion of other relevant sentences. Sentence compression is a recent framework that aims to select the shortest subsequence of words that yields an informative and grammatical sentence. This work proposes a one-step approach for document summarization that jointly performs sentence extraction and compression by solving an integer linear program. We report favorable experimental results on newswire data.

We present a new method that compresses sentences by removing words. In a first stage, it generates candidate compressions by removing branches from the source sentence’s dependency tree using a Maximum Entropy classifier. In a second stage, it chooses the best among the candidate compressions using a Support Vector Machine Regression model. Experimental results show that our method achieves state-of-the-art performance without requiring any manually written rules. 1

Paraphrasing methods recognize, generate, or extract phrases, sentences, or longer natural language expressions that convey almost the same information. Textual entailment methods, on the other hand, recognize, generate, or extract pairs of natural language expressions, such that a human who reads (and trusts) the first element of a pair would most likely infer that the other element is also true. Paraphrasing can be seen as bidirectional textual entailment and methods from the two areas are often similar. Both kinds of methods are useful, at least in principle, in a wide range of natural language processing applications, including question answering, summarization, text generation, and machine translation. We summarize key ideas from the two areas by considering in turn recognition, generation, and extraction methods, also pointing to prominent articles and resources.

We propose a global algorithm for learning entailment relations between predicates. We define a graph structure over predicates that represents entailment relations as directed edges, and use a global transitivity constraint on the graph to learn the optimal set of edges, by formulating the optimization problem as an Integer Linear Program. We motivate this graph with an application that provides a hierarchical summary for a set of propositions that focus on a target concept, and show that our global algorithm improves performance by more than 10 % over baseline algorithms. 1

This work surveys existing evaluation methodologies for the task of sentence compression, identifies their shortcomings, and proposes alternatives. In particular, we examine the problems of evaluating paraphrastic compression and comparing the output of different models. We demonstrate that compression rate is a strong predictor of compression quality and that perceived improvement over other models is often a side effect of producing longer output. 1

We present a substitution-only approach to sentence compression which “tightens ” a sentence by reducing its character length. Replacing phrases with shorter paraphrases yields paraphrastic compressions as short as 60 % of the original length. In support of this task, we introduce a novel technique for re-ranking paraphrases extracted from bilingual corpora. At high compression rates1 paraphrastic compressions outperform a state-of-the-art deletion model in an oracle experiment. For further compression, deleting from oracle paraphrastic compressions preserves more meaning than deletion alone. In either setting, paraphrastic compression shows promise for surpassing deletion-only methods. 1

Previous work has shown that high quality phrasal paraphrases can be extracted from bilingual parallel corpora. However, it is not clear whether bitexts are an appropriate resource for extracting more sophisticated sentential paraphrases, which are more obviously learnable from monolingual parallel corpora. We extend bilingual paraphrase extraction to syntactic paraphrases and demonstrate its ability to learn a variety of general paraphrastic transformations, including passivization, dative shift, and topicalization. We discuss how our model can be adapted to many text generation tasks by augmenting its feature set, development data, and parameter estimation routine. We illustrate this adaptation by using our paraphrase model for the task of sentence compression and achieve results competitive with state-of-the-art compression systems.

Abstract. We address the problem of automatically generating concise factual questions from linguistically complex sentences in reading materials. We discuss semantic and pragmatic issues that appear in complex sentences, and then we present an algorithm for extracting simplified sentences from appositives, subordinate clauses, and other constructions. We conjecture that our method is useful as a preliminary step in a larger question generation process. Experimental results indicate that our method is more suitable for factual question generation applications than an alternative text compression algorithm. 1

Most previous studies on meeting summarization have focused on extractive summarization. In this paper, we investigate if we can apply sentence compression to extractive summaries to generate abstractive summaries. We use different compression algorithms, including integer linear programming with an additional step of filler phrase detection, a noisychannel approach using Markovization formulation of grammar rules, as well as human compressed sentences. Our experiments on the ICSI meeting corpus show that when compared to the abstractive summaries, using sentence compression on the extractive summaries improves their ROUGE scores; however, the best performance is still quite low, suggesting the need of language generation for abstractive summarization. 1