Metasets and opinion mining in new decision support system

Abstract

The paper is devoted to the problem of modeling human attitudes towards imprecise ideas. A metaset is used for representing an imprecise concept and Opinion Mining techniques are applied to build a preference function which reflects someone's attitude towards the idea.The preferences are then evaluated as real numbers for the sake of comparison and selection of the best matching instance. The core of the idea of representing any imprecise concept with a metaset lies in splitting it into a treelike hierarchy of related sub-concepts. The nodes of the tree determine the membership degrees for metaset members and they are natural language terms which also describe reasons for some particular member to satisfy the represented idea. The Opinion Mining allows for automatic gathering and evaluation of opinions from the Internet. The proposed mechanism is applied to solve the problem of selecting the car best matching the imprecise idea of a good car for a lady. This approach can be applied in a decision support systems that helps both marketers and customers.