|
Track: Database Management and SDE
Issam Moghrabi
Lebanese American University
475 Riverside Drive Suite 1846
New York, NY 10115
Telephone: 212-870-2592
Fax: 212-870-2762
E-mail: imoghrbi@lau.edu.lb
Clustering Spatial and Aspatial Data Records in Databases
Defining Issue: Due to generation, addition, manipulation, storage, and management of the large amount of data in the databases associated with GIS applications, efficient ways of storage and clustering of records in data buckets are required to minimize retrieval time. GIS Solution: A suggested clustering technique is statistically analyzed, practically tested, and simulated on ArcView software database facilities to assess effectiveness of the proposed algorithm. Methodology: The suggested clustering method is simple in concept and requires a relatively small number of comparisons to initially cluster a given file. It is based on the specification of a threshold value and the concept of vector space such that a record is inserted in a bucket if the Hamming or 2-norm distance between the new record and its neighbor(s) is less than or equal to the specified threshold value. Our experiments have yielded encouraging results. Software: ArcView + Avenue and link to C++ code
|
|
