Automation of Scenario-Based Schema Matcher Optimization

Schema matchers are used to find related entities in schemas.Automated schema matchers are not infallible, consequently they need to improve on accuracy.Since the accuracy of schema matchers is scenario-dependent, our objective was to define universal methods with which the pre-run optimization of schema matchers for a given scenario is feasible.In this paper, we present our enhanced schema matcher optimization framework which allows the automated, scenario-based optimization of schema matchers.The output of this framework is the recombined schema matcher, which attained 33% average f-measure improvement over the input schema matchers.As part of the framework, we also devised a systematic comparison method for schema matcher components, the Comparative Component Analysis.We propose several performance evaluation bases for the ranking of schema matcher components.