http://www.dagstuhl.de/15295

15. – 16. Juli 2015, Event 15295

Scalable Author Disambiguation for Bibliographic Databases: Project Kick-Off

Organisatoren

Marcel R. Ackermann (Schloss Dagstuhl, DE)

Auskunft zu diesem Event erteilt

Heike Clemens

Dokumente

Externe Homepage

Motivation

The correct attribution of scholarly material to their unambiguous authors ranks among the most critical challenges for digital libraries. More generally, the problem of determining which records in a database refer to the same entities is known as "entity resolution" or "author name disambiguation" and constitutes an important field of research within the discipline of natural language processing.

In a joint project, Schloss Dagstuhl's dblp computer science bibliography and the Zentralblatt MATH (located at FIZ Karlsruhe) aim to begin partnering with the Heidelberg Institute for Theoretical Studies (HITS) to find and implement new and state-of-the-art strategies to overcome the challenges of author identification and disambiguation. Zentralblatt MATH and dblp share the challenges associated with author name disambiguation. Due to their partially overlapping, but also partially disjointed data, a joint effort to identify authors based on the combination of the two data sets appears to be very promising. The Natural Language Processing (NLP) Group at the HITS joins the project by providing its extensive experience with graph-based and network methods for NLP tasks such as co-reference resolution, cross-document co-reference resolution, concept and entity disambiguation.

This is the kick-off meeting of a project in the Leibniz Competition 2015.

Keywords

  • Databases
  • Digital libraries
  • Bibliograhic metadata
  • Named entity recognition
  • Author disambiguation
  • Natural language processing

Dagstuhl's Impact

Bitte informieren Sie uns, wenn eine Veröffentlichung ausgehend von
Ihrem Seminar entsteht. Derartige Veröffentlichungen werden von uns in der Rubrik Dagstuhl's Impact aufgelistet und separat in der Bibliothek präsentiert.