Research On Service Evaluation And Selection Based On Fuzzy Mathematics

With the widespread popularity and development of information services, various kinds ofservices emerge on the Internet which have same functional properties but with different QoS. Howto select services that best meet the requirements of users has become one of the most importantproblems in the research field of Services Computing. However, as the increasing complexity ofusers’ properties and running environments, the accompanied user preferences-based serviceselection has received more and more concentration. User preferences-based service evaluation hasinevitably been one of the hot topics in service evaluation. However, most service evaluationmethods are lack of in-depth analysis for user preferences like fuzziness, uncertainty,incompleteness and dynamic, they have not realized the inherent character of each NFP, nor havethey taken the complex preferences of users on NFPs into consideration.It is difficult to objectivelymeasure QoS, and it affects the accuracy and effectiveness of service evaluation. Therefore, thispaper uses fuzzy mathematics theory to solve calculation and evaluation of QoS, and solvecalculation of user preference weight vector for users’ individual requirements; it realizespreference-based service evaluation. On this basis, the best service meets the actual preference ofthe user.The main work as follows: firstly, according to the different mathematical characters of NFPs,properties are divided into fuzziness property, randomness property and certainty property, differentmathematical methods are used to convert uncertainty values to certainty values; Secondly,considering user preferences, concepts of user satisfaction interval and membership degree of usersatisfaction are proposed, QoS values are evaluated by USMP processing and substituted forevaluation values to reflect user requirements. Thirdly, based on the intuitionistic fuzzy theory, theconcepts and calculation methods of incomplete interval-valued intuitionistic judgment matrice andinterval-valued intuitionistic fuzzy vector are proposed, to resolve the problems of fuzziness,uncertainty and incompleteness for user preferences. Fourthly, average age method together withthe UDIFWA operator are used to cope with the dynamic characteristics of user preference, andweight calculation method based on score function is proposed to calculate preference weightvector of QoS for users. At last, combine results above, the final service evaluation values iscalculated by SAW, the best service is obtained for the user according to the value size. Thesimulation and prototype system demonstrate the feasibility and effectiveness of the proposedmethod, it can better satisfy the user preference and improve the accuracy and effectiveness ofservice evaluation and selection.