Interval Type-2 Fuzzy Multiple Criteria Decision Making Methods And The Application In Recommendation System

With the popularization and application of internet, shopping with computers or phones is rapidly growing in popularity, then Personalized Recommendation Systems (PRS) is necessary for internet and E-commerce. RS can filter irrelevant information for users in the age of information explosion to make users find items that they need. Even more valuable is that PRS can recommend unexpected items for users according to user profile. However, many PRSs are all faceing a reality problem that lack of user ratings, and the user information involved in PRS are expressed by language, thus find a new corresponding relation between users and items instead of user-item rating is necessary.The thesis studys the equivalence relation clustering method under the interval type-2 fuzzy enviroment (IT2 FEC). The advantages of fuzzy equivalence relation clustering mehtod relative to fuzzy C-means are talked about that the former can provide dynamic clustering results for decision makers, which can make up the shortcoming of deciding cluster numbers in advance for the later. The detail computing steps of the new clustering method is stated on the basis of similarity method under IT2 FSs. And the flexibility and superiority of the new clustering algorithm are verified by the case of E-commerce recommending cell-phones under the enviroment of linguistic variables. The calculation of similarities between objects in IT2 FEC is contributed to the discovery of neighbor users in PRS. Then, the thesis researches the application of interval type-2 fuzzy analytic network process (IT2 FANP) on MCDM. On the basis of the ranking method of type-1 fuzzy sets is extended to IT2 FS, the computing steps of the new decision method is elaborated in detail. And the new decision method is applied on an illustration of enterprise technology ability enovation that modeled by linguistic variables. The evaluation results are analyzed which can prove the feasibility and effectiveness of the new decision method. At last, this thesis also applies IT2 FEC and IT2 FANP into the PRS of social network on E-commerce platform. The dynamic clustering results of users and items are calculated, and the corresponding relation between users and items are obtained form the analysis of clustering results, which verified the superiority of IT2 FMCDM on the application of PRS.