A B [formula omitted]-tree based indexing technique for fuzzy numerical data

Abstract

This paper proposes an indexing technique for fuzzy numerical data which increases the performance of query processing when the query involves an atomic possibility measured flexible condition. The proposal is based on a classical indexing mechanism for numerical crisp data, B + -tree, which is implemented in most commercial database management systems (DBMS). This makes the proposed technique a good candidate for integration in a fuzzy DBMS when it is developed as an extension of a crisp DBMS. The efficiency of the proposal is contrasted with another indexing method for similar data and queries, G-tree, which is specifically designed to index multidimensional data. Results show that the proposal performance is similar to and more stable than the measured for G-tree when used for indexing fuzzy numbers.