GRAS: An Adaptive Personalization Scheme for Hypermedia Databases

In this paper we present a new personalisation algorithm for hypermedia databases, called GRAS (Gaussian Rating Adaptation Scheme), which combines content-based and social filtering. The goal is to filter documents retrieved by a query according to the personal interest of the user and to sort them according to their relevance. GRAS is based on ideas about human taste originating in cognitive science and techniques used in marketing. The algorithm makes the benefits of social filtering available to situations where social filtering could not be applied due to lack of a critical mass of users or improper content structure. The algorithm collects background information about the user and the content by implicit and explicit feedback techniques. This information is then used to consecutively adapt userand object profiles using a user model based on cognitive psychology. The algorithm is applicable for the personalisation of any kind of multimedia data and any application domain. GRAS is implemented as a generic personalisation module in the multimedia database MultiMAP1.