Output details
11 - Computer Science and Informatics
University of Sheffield
Disambiguating identity web references using Web 2.0 data and semantics
<16>Disambiguating references to persons on the Web is fundamental to applications like identification and prevention of identity theft. This paper presents a method for this whose novelty is (1) a technique to construct a profile from an individual’s social media presence (e.g. Facebook) (2) two approaches (semi-supervised learning and graph-based random walk) to use the profile for disambiguation. The approach outperforms prior work and approximates human performance. It has led to further projects funded by TSB/Home Office (TRIDS) and the EPSRC (Randms, LODIE) and an application commercialized by our spin-out K-Now. JWS (IF:3.412) is the top semantic web journal.