Second Workshop on Mining Scientific Papers: Computational Linguistics and Bibliometrics

The open access movement in scientific publishing and search engines like Google Scholar have made scientific articles more broadly accessible. During the last decade, the availability of scientific papers in full text has become more and more widespread thanks to the growing number of publications on online platforms such as ArXiv and CiteSeer.
The efforts to provide articles in machine-readable formats and the rise of Open Access publishing have resulted in a number of standardized formats for scientific papers (such as NLM-JATS, TEI, DocBook), full-text datasets for research experiments (PubMed, JSTOR, etc.) and corpora (iSearch, etc.). At the same time, research in the field of Natural Language Processing have provided a number of open source tools for versatile text processing (e.g. NLTK, Mallet, OpenNLP, CoreNLP, Gate, CiteSpace).
Scientific papers are highly structured texts and display specific properties related to their references but also argumentative and rhetorical structure. Recent research in this field has concentrated on the construction of ontologies for citations and scientific articles (e.g. CiTO, LinkedScience) and studies of the distribution of references. However, up to now full-text mining efforts are rarely used to provide data for bibliometric analyses. While bibliometrics traditionally relies on the analysis of metadata of scientific papers (see e.g. a recent special issue on Combining Bibliometrics and Information Retrieval, Mayr & Scharnhorst, 2015), we will explore the ways full-text processing of scientific papers and linguistic analyses can play. With this workshop we like to discuss novel approaches and provide insights into scientific writing that can bring new perspectives to understand both the nature of citations and the nature of scientific articles. The possibility to enrich metadata by the full-text processing of papers offers new fields of application to bibliometrics studies.
Working with full text allows us to go beyond metadata used in bibliometrics. Full text offers a new field of investigation, where the major problems arise around the organization and structure of text, the extraction of information and its representation on the level of metadata. Furthermore, the study of contexts around in-text citations offers new perspectives related to the semantic dimension of citations. The analyses of citation contexts and the semantic categorization of publications will allow us to rethink co-citation networks, bibliographic coupling and other bibliometric techniques.