Linking Citations to their Bibliographic references

Proceedings of the ACL-2012 Special Workshop on Rediscovering 50 Years of Discoveries, pages 110–113, Jeju, Republic of Korea, 10 July 2012. c⃝2012 Association for Computational Linguistics Linking Citations to their Bibliographic references Huy Do Hoang Nhat Web IR / NLP Group (WING) National University of Singapore huydo@comp.nus.edu.sg Praveen Bysani Web IR / NLP Group (WING) National University of Singapore bpraveen@comp.nus.edu.sg Abstract In this paper we describe our participation in the contributed task at ACL Special work- shop 2012. We contribute to the goal of en- riching the textual content of ACL Anthology by identifying the citation contexts in a paper and linking them to their corresponding ref- erences in the bibliography section. We use Parscit, to process the Bibliography of each paper. Pattern matching heuristics are then used to connect the citations with their ref- erences. Furthermore, we prepared a small evaluation dataset, to test the efficiency of our method. We achieved 95% precision and 80% recall on this dataset. 1 Introduction ACL Anthology represents the enduring effort to digitally archive all the publications related to CL and NLP, over the years. Recent work by (Bird et al., 2008) to standardize the corpus in ACL An- thology, makes it more than just a digital reposi- tory of research results. The corpus has metadata information such as ‘title’, ‘author (s)’, ‘publication venue’ and ‘year’ about each paper along with their extracted text content. However it lacks vital in- formation about a scientific article such as position of footnote (s), table (s) and figure captions, biblio- graphic references, italics/emphasized text portions, non-latin scripts, etc. We would like to acknowledge funding support in part by the Global Asia Institute under grant no. GAI-CP/20091116 and from the National Research Foundations grant no. R-252- 000-325-279. The special workshop at ACL 2012, celebrates 50 years of ACL legacy by gathering contributions about the history, evolution and future of compu- tational linguistics. Apart from the technical pro- gramme, the workshop also hosts a contributed task to enrich the current state of Anthology corpus. A rich-text format of the corpus will serve as a source of study for research applications like citation anal- ysis, summarization, argumentative zoning among many others. We contribute to this effort of enriching the An- thology, by providing a means to link citations in an article to their corresponding bibliographic refer- ences. Robert Dale 1 defines citation, as a text string in the document body that points to a reference at the end of the document. Several citations may co-refer to a single reference string. As an example consider the following sentence, Few approaches to parsing have tried to handle disfluent utterances (notable exceptions are Core & Schubert, 1999; Hindle, 1983; Nakatani & Hirschberg, 1994). The portion of texts in italics are the citations and we intend to annotate each citation with an unique identifier of their bibliographic reference. <ref target="BI10">Hindle, 1983</ref> Such annotations are useful for navigating be- tween research articles and creating citation net- works among them. These networks can be used …

Content truncated to 500 words