Using Citations to Generate surveys of Scientific Paradigms

Human Language Technologies: The 2009 Annual Conference of the North American Chapter of the ACL, pages 584–592, Boulder, Colorado, June 2009. c⃝2009 Association for Computational Linguistics Using Citations to Generate Surveys of Scientific Paradigms Saif Mohammad†∗, Bonnie Dorr†‡∗, Melissa Egan†‡, Ahmed Hassanφ, Pradeep Muthukrishanφ, Vahed Qazvinianφ, Dragomir Radevφ§, David Zajic‡⋄ Institute for Advanced Computer Studies† and Computer Science‡, University of Maryland. Human Language Technology Center of Excellence∗. Center for Advanced Study of Language⋄. {saif,bonnie,mkegan,dmzajic}@umiacs.umd.edu Department of Electrical Engineering and Computer Scienceφ School of Information§, University of Michigan. {hassanam,mpradeep,vahed,radev}@umich.edu Abstract The number of research publications in var- ious disciplines is growing exponentially. Researchers and scientists are increasingly finding themselves in the position of having to quickly understand large amounts of tech- nical material. In this paper we present the first steps in producing an automatically gen- erated, readily consumable, technical survey. Specifically we explore the combination of citation information and summarization tech- niques. Even though prior work (Teufel et al., 2006) argues that citation text is unsuitable for summarization, we show that in the frame- work of multi-document survey creation, cita- tion texts can play a crucial role. 1 Introduction In today’s rapidly expanding disciplines, scientists and scholars are constantly faced with the daunting task of keeping up with knowledge in their field. In addition, the increasingly interconnected nature of real-world tasks often requires experts in one dis- cipline to rapidly learn about other areas in a short amount of time. Cross-disciplinary research requires scientists in areas such as linguistics, biology, and sociology to learn about computational approaches and appli- cations, e.g., computational linguistics, biological modeling, social networks. Authors of journal ar- ticles and books must write accurate surveys of pre- vious work, ranging from short summaries of related research to in-depth historical notes. Interdisciplinary review panels are often called upon to review proposals in a wide range of areas, some of which may be unfamiliar to panelists. Thus, they must learn about a new discipline “on the fly” in order to relate their own expertise to the proposal. Our goal is to effectively serve these needs by combining two currently available technologies: (1) bibliometric lexical link mining that exploits the structure of citations and relations among citations; and (2) summarization techniques that exploit the content of the material in both the citing and cited papers. It is generally agreed upon that manually written abstracts are good summaries of individual papers. More recently, Qazvinian and Radev (2008) argue that citation texts are useful in creating a summary of the important contributions of a research paper. The citation text of a target paper is the set of sen- tences in other technical papers that explicitly refer to it (Elkiss et al., 2008a). However, Teufel (2005) argues that using citation text directly is not suitable for document summarization. In this paper, we compare and contrast the use- fulness of abstracts and of citation text in automati- cally generating a technical survey on a given topic from multiple research papers. The next section pro- …

Content truncated to 500 words