A schema matching system for on-the-fly autonomous data integration

Abstract

Conventional schema matching systems usually choose a randomly ordered set of simple matchers and subsequently combine their individual scores with the help of a composite function ignoring the properties of the individual matchers and disregarding the context of the application of the matchers and therefore, culminate counterintuitive scores, improper matches, wasteful and counterproductive computations. Moreover, as users can hardly validate the processing accuracy until the computation is complete, matching efficiency and quality assurance play a crucial role in autonomous systems. In this paper, we propose a new method, OntoMatch, for schema matching that can avoid wasteful computation by a prudent and objective selection of the ordering of a subset of useful matchers and consequently improves the matching efficiency and accuracy. Experimental results corroborate that successive applications of the simple matchers in OntoMatch monotonically improve the matching score.