Towards Browsing Distant Metadata with Semantic Signature

Candidate: Andrew Shek-Ting Choi
Type: Master of Science (MSc), School of Interactive Arts and Technology
Date: August 3, 2005
Senior Supervisor: Dr. Marek Hatala, Associate Professor
Thesis: Download Thesis Document

Abstract

One of the benefits of E-Learning network is to connect users to distributed learning repositories where they can expose to numerous learning resources. However, metadata of learning resources resided in different repositories are very often annotated with concepts defined by different ontology specific to the organization. That makes finding information based on local conceptual framework difficult. Different organization with different background and target audience may use different terms with similar semantic to define and describe two similar learning resources. As such, using keyword approach to find relevant information may not yield satisfactory result.

In this thesis, we describe a light-weighted information integration solution for browsing federally distributed metadata without incurring expensive schema matching or data mapping. I present experiments on real-world data that validate the proposed solution. Finally, I will discuss how this approach can simplify semantic mapping and enhance browsing experience in a cooperative learning network.