Machine Learning Methods for Ontology Mining
Nicola Fanizzi, Claudia d’Amato, Floriana Esposito · 2010
While machine learning and data-mining methods have been developed for dealing with traditional database settings (possibly extended to the multirelational setting), in the vision of the semantic Web new forms of distributed knowledge bases (ontologies) have been established that require an appropriate treatment and novel specific solutions. The challenge is represented by the natural open-world semantics of the ontologies in a context where new resources (and services) may continuously be made available. Solutions are required for new interesting problems that are related to the semiautomation of various knowledge-intensive tasks, such as ontology building, population, mapping, integration, query answering, and many others. The chapter surveys machine learning techniques that have been applied to solve ontology mining problems since the early 1990s. Specifically, given the standard formal representations proposed, ultimately based on description logics, the presented algorithms have been devised to perform classical data-mining tasks in the new semantic framework.