Supporting Failing Database Queries in a Flexible Context: A Data-Driven Approach

Abstract

We investigate the problem of handling of failing queries involving fuzzy predicates. We propose an approach that leverages data distribution of the target database. It consists in two steps: i) Query translation that aims at translating the failing fuzzy query into a crisp query by means of a particular semantic distance between sets; ii) Query relaxation which consists in expanding the translated query criteria with similar values. To rank-order the approximate query results, a method is proposed