Extending a Single-Document Summarizer to Multi-Document: a Hierarchical Approach

Proceedings of the Fourth Joint Conference on Lexical and Computational Semantics (*SEM 2015), pages 176–181, Denver, Colorado, June 4–5, 2015. Extending a Single-Document Summarizer to Multi-Document: a Hierarchical Approach Lu´ıs Marujo1,2,3, Ricardo Ribeiro1,4, David Martins de Matos1,2, Jo˜ao P. Neto1,2, Anatole Gershman3, and Jaime Carbonell3 1INESC-ID Lisboa, 2IST/ULisboa, 4ISCTE-IUL, Lisboa, Portugal 3School of Computer Science, CMU, Pittsburgh, USA {lmarujo,anatoleg,jgc}@cs.cmu.edu {ricardo.ribeiro,david.matos,joao.neto}@inesc-id.pt Abstract The increasing amount of online content mo- tivated the development of multi-document summarization methods. In this work, we explore straightforward approaches to extend single-document summarization methods to multi-document summarization. The pro- posed methods are based on the hierarchical combination of single-document summaries, and achieves state of the art results. 1 Introduction The use of the Internet to fulfill generic informa- tion needs motivated pioneer multi-document sum- marization efforts as NewsInEssence (Radev et al., 2005) or Newsblaster (McKeown et al., 2002), on- line since 2001. In general, multi-document sum- marization approaches have to address two differ- ent problems: passage selection and information or- dering. Current multi-document systems adopt, for passage selection, approaches similar to the ones used in single-document summarization, and use the chronological order of the documents for informa- tion ordering (Christensen et al., 2013). The prob- lem is that most approaches fail to generate sum- maries that cover generic topics which comprehend different, equally important, subtopics. We propose to extend a state-of-the-art single-document summarization method, KP- CENTRALITY (Ribeiro et al., 2013), capable of focusing on diverse important topics while ignoring unimportant ones, to perform multi-document sum- marization. We explore two hierarchical strategies to perform this extension. This document is organized as follows: Sect. 2 ad- dresses the related work; Sect. 3 presents our multi- document summarization appproach; experimental results close the paper. 2 Related Work Most of the current work in automatic summariza- tion focuses on extractive summarization. The most popular baselines for multi-document summariza- tion fall into one of the following general mod- els: Centrality-based (Radev et al., 2004; Erkan and Radev, 2004; Wang et al., 2008; Ribeiro and de Matos, 2011), Maximal Marginal Relevance (MMR) (Carbonell and Goldstein, 1998; Guo and Sanner, 2010; Sanner et al., 2011; Lim et al., 2012), and Coverage-base methods (Lin and Hovy, 2000; Sipos et al., 2012). Additionally, methods such as KP-CENTRALITY (Ribeiro et al., 2013), which is centrality and coverage-based, follow more than one paradigm. In general, Centrality-based models are used to produce generic summaries, while the MMR family generates query-oriented ones. Coverage- base models produce summaries driven by words, topics or events. Centrality-as-relevance methods base the detec- tion of the most salient passages on the identification of the central passages of the input source(s). One of the main representatives of this family is Passage- to-Centroid Similarity-based Centrality. Centroid- based methods build on the idea of a pseudo-passage that represents the central topic of the input source— the centroid—selecting as passages to be included in the summary the ones that are close to the centroid. Another approach to centrality estimation is to com- 176 pare each candidate passage to every …

Content truncated to 500 words