Đang chuẩn bị nút TẢI XUỐNG, xin hãy chờ
Tải xuống
Case-based reasoning (CBR) and rule-based reasoning (RBR) are two important methods in the knowledge engineering to support decision making (see [2, 7, 8, and 9]) in decision support systems (DSS). In rule-based reasoning the computer examines historical cases and generates rules, which are chained (forward or backward) to solve problems. Case-based reasoning, on the other hand, follows a different process, it finds those cases in memory that shoved problems similar to the current problem, and then adapts the previous solutions to fit the current problem by taking in to account any difference between the current and previous situations. As the main obstacle of RBR method of generating. | CASE-BASED REASONING WITH ROUGH FEATURES Hoang Xuan Huan 1. Introduction Case-based reasoning CBR and rule-based reasoning RBR are two important methods in the knowledge engineering to support decision making see 2 7 8 and 9 in decision support systems DSS . In rule-based reasoning the computer examines historical cases and generates rules which are chained forward or backward to solve problems. Case-based reasoning on the other hand follows a different process it finds those cases in memory that shoved problems similar to the current problem and then adapts the previous solutions to fit the current problem by taking in to account any difference between the current and previous situations. As the main obstacle of RBR method of generating rules from experiments limited its capacity of application then the CBR proved to be an extremely effective approach in complex case see 9 . In order to find similar cases we have to assign appropriate features to each case. In many problems a feature value may be uncertainly determined. For example a number feature r may belong to an interval a b and its distribution can not be known. Likes other areas see 5 this situation demands us to have another approach to extend the scope of application. Basing on rough set theory 6 and interval algebra 1 Lingras 3 4 proposed the concept of rough patterns which are based on the notion of rough numbers called rough values . A rough number consists of an upper and a lower bound which can be used to represent a range of values for variables such that daily temperature or daily financial indicators. By extending the Lingras s notion of rough values in this paper we propose an approach to use CBR in the case that the features are determined uncertainly. Except the conclusion section this paper is organized as follows. The basic notions and characters of DSS and CBR are briefly presented in section 2. A model of CBR with rough features and an example of its application to health care are presented
