[AISWorld] RefSeer: Citation Recommendation System

[Lee Giles]

Apologies for any cross-postings.

SYSTEM BETA RELEASE

===================

REFSEER: Citation Recommendation based on Topic Modeling
(http://refseer.ist.psu.edu)

DESCRIPTION: RefSeer (http://refseer.ist.psu.edu), a beta release Citation
Recommendation System, is hosted at the Pennsylvania State University.
Given either text from an abstract/description or a pdf document of a paper
or part of a paper, RefSeer recommends documents from the CiteSeerX repository
as citations. RefSeer internally computes from the text a topical composition
based on topic modeling. Based upon this composition, recommended citations for
majority topics are ranked and shown in a faceted interface. Refseer uses documents
in CiteSeerX for both its offline training and online recommendation. It is
currently trained with over 1.5 million documents and new documents
are added monthly in its citation database.

CONTACT: Saurabh Kataria (skataria at ist.psu.edu); Prasenjit Mitra
(pmitra at ist.psu.edu); C. Lee Giles (giles at ist.psu.edu)