Identifying Non-Explicit Citing Sentences for Citation-Based Summarization.

Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics, pages 555–564, Uppsala, Sweden, 11-16 July 2010. c⃝2010 Association for Computational Linguistics Identifying Non-explicit Citing Sentences for Citation-based Summarization Vahed Qazvinian Department of EECS University of Michigan Ann Arbor, MI vahed@umich.edu Dragomir R. Radev Department of EECS and School of Information University of Michigan Ann Arbor, MI radev@umich.edu Abstract Identifying background (context) informa- tion in scientific articles can help schol- ars understand major contributions in their research area more easily. In this paper, we propose a general framework based on probabilistic inference to extract such context information from scientific papers. We model the sentences in an article and their lexical similarities as a Markov Ran- dom Field tuned to detect the patterns that context data create, and employ a Belief Propagation mechanism to detect likely context sentences. We also address the problem of generating surveys of scien- tific papers. Our experiments show greater pyramid scores for surveys generated us- ing such context information rather than citation sentences alone. 1 Introduction In scientific literature, scholars use citations to re- fer to external sources. These secondary sources are essential in comprehending the new research. Previous work has shown the importance of cita- tions in scientific domains and indicated that ci- tations include survey-worthy information (Sid- dharthan and Teufel, 2007; Elkiss et al., 2008; Qazvinian and Radev, 2008; Mohammad et al., 2009; Mei and Zhai, 2008). A citation to a paper in a scientific article may contain explicit information about the cited re- search. The following example is an excerpt from a CoNLL paper1 that contains information about Eisner’s work on bottom-up parsers and the notion of span in parsing: “Another use of bottom-up is due to Eisner (1996), who introduced the notion of a span.” 1Buchholz and Marsi “CoNLL-X Shared Task On Multi- lingual Dependency Parsing”, CoNLL 2006 However, the citation to a paper may not always include explicit information about the cited paper: “This approach is one of those described in Eis- ner (1996)” Although this sentence alone does not provide any information about the cited paper, it suggests that its surrounding sentences describe the proposed approach in Eisner’s paper: “... In an all pairs approach, every possible pair of two tokens in a sentence is considered and some score is assigned to the possibility of this pair having a (directed) dependency rela- tion. Using that information as building blocks, the parser then searches for the best parse for the sentence. This approach is one of those de- scribed in Eisner (1996).” We refer to such implicit citations that contain information about a specific secondary source but do not explicitly cite it, as sentences with con- text information or context sentences for short. We look at the patterns that such sentences cre- ate and observe that context sentences occur with- ing a small neighborhood of explicit citations. We also discuss the problem of extracting context sen- tences for a source-reference article pair. We pro- pose a general framework that …

Content truncated to 500 words