(An) integrated cooperative query answering approach based on data abstraction and fuzzy relation

Abstract

Although a query language has been used as a convenient tool to obtain information from a database, users increasingly demand more intelligent cooperative query answering approaches that can understand the intent of an imprecise query and provide additional useful information as well as exact answers. However, existing approaches have limitations in the diversity of admitted queries, query relaxation control through user interaction, and knowledge maintenance because they cannot extract enough semantic knowledge from underlying database to support effective information retrieval and decision support. To remedy such shortcomings, this dissertation proposes an integrated cooperative query answering approach that relaxes a search scope of the original query condition and provides approximate neighborhood answers as well as exact answers. The main issues on cooperative query answering are the representation of similarity relationships between data values, which are necessary for query relaxation and inference support between database objects. The proposed knowledge representation framework, fuzzy abstraction hierarchy (FAH), adopts abstraction hierarchy and fuzzy relation approaches to extract the similarity relationship between data values. FAH is a multilevel knowledge representation framework that extracts abstract values from an underlying database and puts them into abstraction hierarchies. Additionally, it adopts fuzzy relation to represent discriminated similarity strength between data values. Thus, the similarity measure between values in FAH is represented by the notion of data abstraction and fuzzy relation. As a result, FAH acquires richer semantic knowledge of an underlying database and embraces the advantages of existing data abstraction and semantic distance approaches. Based on FAH, query relaxation operators such as query generalization, approximation, and specialization of a value are developed. The semantic knowledge involved in FAH is incorpora...