Đang chuẩn bị liên kết để tải về tài liệu:
Hedges algebras and fuzzy partition problem for qualitative attributes

Không đóng trình duyệt đến khi xuất hiện nút TẢI XUỐNG Tải xuống

In this paper, the author proposed a new method to construct the membership functions (MFs) based on database. The theory of hedge algrebra was used to build the membership functions and GA is applied to optimize them. The experimental results demonstrate the benefits of this method. | Journal of Computer Science and Cybernetics, V.32, N.4 (2016), 335–350 DOI 10.15625/1813-9663/32/4/9145 HEDGES ALGEBRAS AND FUZZY PARTITION PROBLEM FOR QUALITATIVE ATTRIBUTES TRAN THAI SON1 , NGUYEN TUAN ANH2 1 Institute of Information Technology, Vietnam Academy of Science and Technology 2 University of Information and Communication Technology, Thai Nguyen University 1 ttson1955@gmail.com; 2 anhnt@ictu.edu.vn Abstract. There have been many approaches proposed to derive membership functions for mining fuzzy association rules using genetic algorithms (GAs) and show their advantages. However, these approaches show that the number of linguistic terms needs to be predefined. In this paper, the author proposed a new method to construct the membership functions (MFs) based on database. The theory of hedge algrebra was used to build the membership functions and GA is applied to optimize them. The experimental results demonstrate the benefits of this method. Keywords. Fuzzy Association Rules; Data mining; Hedge algebras; Genetic algorithms; Membership functions 1. INTRODUCTION Recently, mining fuzzy association rules, such as “If students have high academic results and are passionate about researching, they will find a good job.”, has been the topic that is considered and developed. To be able to get fuzzy rules, in the first step, we partition each attribute domain of a database into fuzzy sets, characterized by membership functions (MFs). Then, each value (number) in the database will be converted to the corresponding set of the degree of membership. Numerical Database is converted into a fuzzy database ready for fuzzy mining in next steps. Traditionally, the fuzzy database is often assumed to be available, i.e, the MF sets are defined for each DB attribute. Such MF sets are usually determined experimentally and independently from database. In fact, it was shown that the approach may inversely affect the quality of the mining rules. Therefore, researchers .