In this research seminar recorded at Talis in May 2010, Matthew Rowe describes his Ph.D. research on Disambiguating Identity Web References using Social Data.

(See this video on Vimeo for alternative media players.)

Matt gives the following background to his research:

“The availability of personal information on the Web has lead to a rise in malevolent practices such as identity theft and lateral surveillance, this in turn has forced web users to find and disambiguate web resources containing their personal information. I will present three distinct automated disambiguation techniques designed to overcome the burden placed on web users, each of these techniques fuse Semantic Web technologies with existing state of the art techniques (inference rules, graph-based random walks and machine learning). Social data leveraged from Social Web platforms (Facebook, MySpace and Twitter) is used to support the disambiguation process and a thorough evaluation of the presented techniques demonstrates their ability to outperform humans at the same task.”

At the time of the talk, Matt was a Ph.D. student within the University of Sheffield’s Organisations, Information and Knowledge (OAK) Group. His thesis, successfully defended just days after this seminar was given, explored automated disambiguation techniques supported by data leveraged from Social Web platforms.

His research has drawn on several different areas of Computer Science including the Semantic Web, the Social Web and machine learning. Recent research efforts have centred around the Linked Data field, in particular exploring how URI disambiguation can be tackled from a statistical, machine learning perspective. Matt is now a Research Associate at The Open University’s Knowledge Media Institute.

Links:

• Matt Rowe’s Home Page
• Ph.D. Thesis – Disambiguating Identity Web References using Social Data