Exploiting Social Annotations for Resource Classification

Abstract

The lack of representative textual content in many resources suggests the study of additional metadata to improve classification tasks. Social bookmarking and cataloging sites provide an accessible way to increase available metadata in large amounts with user-provided annotations. In this chapter, the authors study and analyze the usefulness of social annotations for resource classification. They show that social annotations outperform classical content-based approaches, and that the aggregation of user annotations creates a great deal of meaningful metadata for this task. The authors also present a method to get the most out of the studied data sources using classifier committees.