Group users service selection based on prediction

Presently, web service selection is mainly based on QoS attributes, and decision making of service selection is mainly a single user. The QoS attribute values are assumed to be accurate and complete. But in practice, some QoS attribute values are often missing or wrong. This situation brings a great challenge to the service selection. Moreover, the subject of service selection is often the group users. In the course of group user service selection based on QoS attribute values, conflicts often occur. In order to solve the problem of some service attribute values are missing and group user service selection, we take the following measures. Firstly, according to the users´ previous information of service selection, we use neural network to predict the missed service QoS attribute values. Secondly, we improve the TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) algorithm and apply it to resolve the group user service selection problem. Previously, the TOPSIS algorithm is mainly applied to single user decision-making. We have improved this algorithm and apply it for group users decision-making. This algorithm provides the group users highest satisfaction degree, and avoids the conflict problem in group service selection.