Personalized News Search in WWW: Adapting on User's Behavior

Abstract

Personalized Web Search becomes nowadays a promising option in the field of Information Retrieval and search engines design by improving both output quality and user experience. In this paper, we present and evaluate the subsystem, which conducts the Advanced and Personalized search of PeRSSonal, a web-based mechanism for the retrieval, processing and presentation of articles and RSS feeds collected from major news portals of the Internet. The proposed technique uses information explicitly provided by the user in his profile as well as information that the mechanism can learn from the user’s behavior during his search and browsing sessions in the system. As this behavior dynamically evolves, the same happens to the user’s interests under the prism of the search engine. By adopting this user-centric approach, we manage to present the user with better-refined and more focused results, incorporating his personal preferences to the output. The algorithm operates not in a stand-alone manner but it co-operates and binds with the rest of modules of PeRSSonal in order to accomplish maximum integration with the system. Furthermore, we introduce an enhancement in the search function, based on cached results from past search sessions of each user individually.