The ability of the web site to automatically reorganize the site structure towards the benefit of user groups or individuals, known as web site adaptation, is becoming a major requirement for web site designers to improve their sites' usability. Clustering techniques are used as the basis of many existing approaches for developing adaptive web sites. However the used clustering techniques limit the efficiency of these approaches. We developed a new soft clustering technique to be used as the underlying technique for adaptive web sites. In particular, it proposes the use of soft k-medoids technique to provide a more efficient solution to the index page synthesis problem. The performance of the proposed systems is evaluated using real data. The performance evaluation experiments showed the effectiveness of the proposed systems. Also, the experiments showed the scalability of the proposed systems for an increasing number of users and for an increasing size of web logs.