Research On Service Selection Based On User Risk Preferences

With the advances in network and information technology, service computing technology attract people’s attention gradually. Service selection is one of the hot issues of current service computing. On one hand, with the explosive growth in the number of services, redundant service attribute information will reduce the efficiency of service selection. Attribute reduction before services selection can effectively reduce the redundant information contained in the service selection process. In reduction process, different users have different understanding attitudes towards effective attributes on the same data table. Attribute reduction on interval attribute only based on dominance relations will make significant departures from the user’s needs. On the other hand, service attributes rating contains only part of the user’s rating information. The loss of rating information will cause the volatility of the service quality. The volatility of the service quality can lead to the uncertainty of service results, and then users select a service at risk, so there is a deviation between the expected value and the actual selection result. Users’ attitudes toward risk are different, some users want to take less risk to get a relatively safe service, some user want to take more risks in order to achieve greater likelihood of service revenue, so user’s risk preference could largely affect the selection results. To solve the above problems, this paper proposes a service selecction method based on risk preference according to utility theory and perceived risk theory, the main work includes the following aspects:1)Build rating range to make a complete description of services non-functional attributes values.The description contains the missing imformation and service selection risk posed by missing information. An improved method of attribute reduction based on dominance relations is proposed to remove redundant information. The method first divides users into groups according to the users’ risk preference, then formulates expected rules to improve attribute reduction based on dominance relations,and finally the attributes sets are reduced and reduction results are closer to the user’s needs.2)The loss of rating information will cause fluctuations in the overall quality of service. This paper proposes a service selection method based on users’ risk preferences to solve the risk problem caused by service quality volatility. The method applys utility theory and user’s perceived risk into the service selection process to improve traditional service selection method based on Qo S, which contributes to express users’ risk preference according to Qo S rating. Meanwhile, attribute preference information is contained to improve user’s personality characteristics to make the selection results closer to the user’s needs.3) The paper optimizes users’ needs further, and designs a service selection prototype and demonstration system based on users’ risk preference. System implementation is analyzed in details in the process of demand analysis, general design and module design to illustrate effectiveness of the method proposed in this paper.