Mining ontology for automatically acquiring Web user information needs

It is not easy to obtain the right information from the Web for a particular Web user or a group of users due to the obstacle of automatically acquiring Web user profiles. The current techniques do not provide satisfactory structures for mining Web user profiles. This paper presents a novel approach for this problem. The objective of the approach is to automatically discover ontologies from data sets in order to build complete concept models for Web user information needs. It also proposes a method for capturing evolving patterns to refine discovered ontologies. In addition, the process of assessing relevance in ontology is established. This paper provides both theoretical and experimental evaluations for the approach. The experimental results show that all objectives we expect for the approach are achievable.

1 Introduction

OVER the last decade, we have witnessed an explosive growth in the data accessible on the Web. However, there are two fundamental issues regarding the effectiveness of information gathering from the Web: mismatch and overload. Mismatch means some useful and interesting data has been overlooked, whereas overload means some gathered data is not what users want.