Effective Quality Of Service Browsing For Web Service Selection

The growing number of Services on the Web has made locating desired Web Services a sizable challenge. Web Service requestors deem a Quality of Service (QoS) based Web Service selection important in terms of providing a relevant and user centric service selection experience. In this thesis an interactive QoS based Web Service browsing mechanism is proposed, which makes use of three clustering algorithms including vector-based, preference-based and weighted clustering. We use symbolic interval data as the principle representation of QoS attributes. The browsing mechanism which was implemented as part of this research allows service requestors to prioritize their search by hierarchically clustering their web services. This is done in order of their preferences and also by attaching a weight to each QoS attribute, which is a beneficial compromise between performance-high preference-based clustering and time-efficient vector-based clustering, Along with several extensive experiments, a user study was conducted in order to test the usability of this browsing mechanism and to test the overall efficiency and performance of the three clustering algorithms in comparison. The result of the experiment led to evidences that preference-based browsing approach was the most efficient one when compared to vector-based or weighted clustering approaches.