Extraction of Synonymies, Homonymies, and Type Conflicts

Abstract

This chapter is devoted to the illustration of algorithms for extracting synonymies, homonymies and type conflicts. The plan of the chapter is as follows: first we introduce SD-Graphs, graph-based structures which can be associated to database schemes; then we define a metrics associated to SD-Graphs which allows us to measure the semantic distance existing between objects, which is central to support the derivation of interscheme properties. After that the metrics has been introduced we present a framework managing both the derivation of synonymies, homonymies and type conflicts as well as the resolution of these latter ones. Then, we illustrate a complete example showing the behaviour of the presented algorithms. The last two sections of this chapter are devoted to the computation of the complexity of the presented algorithms and to the illustration of a sensitivity analysis carried out on them. In particular, as far as the complexity is concerned, we show that the presented algorithms are polynomial in the size of input schemes. As for the sensitivity analysis, we prove that the quality of the results obtained by the algorithms is not affected by possible (little) errors occurring either in the basic information provided by human experts or in thresholds and weights.

The material presented in this chapter is taken from [92,122,87,90,117,118].