Fuzzy query processing using clustering techniques

• Published in: Information processing & management Vol. 26; no. 2; pp. 279 - 293
• Main Authors: Kamel, M., Hadfield, B., Ismail, M.
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 1990; Elsevier Science; Pergamon Press; Elsevier Science Ltd
• Subjects: Architecture; Clustering; Computerised information retrieval; Computerized information storage and retrieval; Exact sciences and technology; Implementations; Information and communication sciences; Information and document management systems; Information processing; Information processing and retrieval; Information retrieval; Information retrieval systems. Information and document management system; Information science. Documentation; Information storage and retrieval; Information work; Mathematical models; Miscellaneous; Online information retrieval; Prototypes; Sciences and techniques of general use; Searching; Subject indexing; Technical services; Fuzzy query; Document retrieval; Documentation data processing; Processor; Prototype; Query; High level language; Computerized processing; Data processing; Fuzzy set; Algorithm
• ISSN: 0306-4573; 1873-5371
• DOI: 10.1016/0306-4573(90)90031-V

This paper addresses the problem of processsing fuzzy queries in databases and information retrieval systems. Most of the existing approaches for handling fuzziness in queries require explicit definitions of fuzziness and membership functions. We propose an architecture and data structures for a fuzzy query processor that utilizes clustering techniques as a tool to generate the mapping between fuzzy terms, defined at a high level of abstraction and the data items of the database records. The clustering techniques developed in this paper are based on multiple thresholding of fuzzy clustering. The use of thresholded fuzzy clustering provides a controlled overlap between clusters of records and thus reflects, naturally, the required fuzziness in the response. A prototype fuzzy query processor based on this approach has been implemented and tested on a sample database.