Scientific Article Summarization Using Citation-Context and Article’s Discourse Structure

Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 390–400, Lisbon, Portugal, 17-21 September 2015. c⃝2015 Association for Computational Linguistics. Scientific Article Summarization Using Citation-Context and Article’s Discourse Structure Arman Cohan Information Retrieval Lab Computer Science Department Georgetown University arman@ir.cs.georgetown.edu Nazli Goharian Information Retrieval Lab Computer Science Department Georgetown University nazli@ir.cs.georgetown.edu Abstract We propose a summarization approach for scientific articles which takes advantage of citation-context and the document discourse model. While citations have been previously used in generating scientific summaries, they lack the related context from the referenced article and therefore do not accurately reflect the article’s content. Our method overcomes the problem of inconsistency between the citation summary and the article’s content by providing context for each citation. We also leverage the inherent scientific article’s discourse for producing better summaries. We show that our proposed method effectively improves over existing summarization approaches (greater than 30% improvement over the best performing baseline) in terms of ROUGE scores on TAC2014 scientific summarization dataset. While the dataset we use for evaluation is in the biomedical domain, most of our approaches are general and therefore adaptable to other domains. 1 Introduction Due to the expanding rate at which articles are being published in each scientific field, it has become difficult for researchers to keep up with the developments in their respective fields. Scientific summarization aims to facilitate this problem by providing readers with concise and informative representation of contributions or findings of an article. Scientific summarization is different than general summarization in three main aspects (Teufel and Moens, 2002). First, the length of scientific papers are usually much longer than general articles (e.g newswire). Second, in scientific summarization, the goal is typically to provide a technical summary of the paper which includes important findings, contributions or impacts of a paper to the community. Finally, scientific papers follow a natural discourse. common organization for scientific paper is theA one in which the problem is first introduced and is followed by the description of hypotheses, methods, experiments, findings and finally results and implications. Scientific summarization was recently further motivated by TAC2014 biomedical summarization track1 in which they planned to investigate this problem in the domain of biomedical science. There are currently two types of approaches towards scientific summarization. First is the articles’ abstracts. While abstracts provide a general overview of the paper, they cannot be considered as an accurate scientific summary by themselves. That is due to the fact that not all the contributions and impacts of the paper are included in the abstract (Elkiss et al., 2008). In addition, the stated contributions are those that the authors deem important while they might be less important to the scientific community. Moreover, contributions are stated in a general and less focused fashion. These problems motivated the other form of scientific summaries, i.e., citation based summaries. Citation based summary is a summary which is formed by utilizing a set of citations to a referenced article (Qazvinian and Radev, 2008; Qazvinian et al., 2013). …

Content truncated to 500 words