![]() | Only 14 pages are availabe for public view |
Abstract In this thesis, a new CBR indexing scheme that uses relational databases is presented The former main idea IS to retrieve cases from database and then apply some processing to them to build the case base index tree. The new indexing scheme attempts to overcome the weak points in the old scheme by building the index in the database and cache this index in the memory. This scheme has many advantages. First, no need for domain knowledge to build the index Second, no loss of related cases. Third, it is a generic scheme that does not need any changes when adding new indexing features or removing old ones. A performance comparative study is included to show that there is a range in which the new scheme is better in terms of retrieval time than the old scheme due to the index building time which is relatively large compared to the retrieval time in both schemes |