Fuzzy queries of numerical attributes for keyword-based search over relational databases

Precision queries of keyword search developed quickly over relational databases, but it can't be better to process fuzzy queries for satisfying higher requests of users. Aiming at fuzzy queries of numerical attributes for keyword-based search over relational databases, we give a new kind of membership function (normal school fuzzy sets). Given several instances of fuzzy queries on merchandise information databases, we introduce and define membership functions to adjust fuzzy scopes. The top-k algorithm is employed to experiment for comparing the two kinds of membership functions in commodity information databases. Experiment results show that average precision and average recall are excellent to correspond to actual situation when query sets are come from normal school fuzzy sets.