Resources and users in the tagging process: approaches and case studies

Abstract

In this contribution we propose a comparison between two distinct approaches to the annotation of digital resources. The former, top-down, is rooted in the cathedral model and is based on an authoritative, centralized defnition of the adopted mark-up language; the latter, bottom-up, refers to the bazaar model and is based on the contributions of a community of users. These two approaches are analyzed taking into account both their descriptive potential and the constraints they impose on the reasoning process of recommender systems, with special reference to user profling. Three case studies are described, with reference to research projects that apply these approaches in the contexts of e-learning and knowledge management.

Keywords

• social tagging
• ontologies
• user profling
• clustering
• neighbour selection