Identifying and eliminating inconsistencies in mappings across hierarchical ontologies

Abstract

Many applications require the establishment of mappings between ontologies. Such mappings are established by domain experts or automated tools. Errors in mappings can introduce inconsistencies in the resulting combined ontology. We consider the problem of identifying the largest consistent subset of mappings in hierarchical ontologies. We consider mappings that assert that a concept in one ontology is a subconcept, superconcept, or equivalent concept of a concept in another ontology and show that even in this simple setting, the task of identifying the largest consistent subset is NP-hard. We explore several polynomial time algorithms for finding suboptimal solutions including a heuristic algorithm to this problem. We experimentally compare the algorithms using several synthetic as well as real-world ontologies and mappings. © 2010 Springer-Verlag.