Correlation-Based Web Document Clustering for Adaptive Web Interface Design

摘要: A great challenge for web site designers is how to ensure users' easy access important pages efficiently. In this paper we present a clustering-based approach address problem. Our perform efficient and effective correlation analysis based on logs construct clusters of reflect the co-visit behavior users. We novel adapting previous clustering algorithms that are designed databases in problem domain page clustering, show our new methods can generate high-quality very large when fail. Based results, then apply data-mined knowledge interfaces improve performance. develop an automatic method interface adaptation: by introducing index minimize overall user browsing costs. The aimed at providing short cuts users get their objective fast, solve previously open determine optimal number pages. empirically performs better than many experiments several realistic log files.