Harvesting Ontologies from Open Domain Corpora : a Dynamic Approach

In this work we present a robust approach for dynamically harvesting domain knowledge from open domain corpora and lexical resources. It relies on the notion of Semantic Domains and provides a fully unsupervised method for terminology extraction and ontology learning. It makes use of an algorithm based on Conceptual Density to extract useful relations from WordNet. The method is efficient, accurate and widely applicable, as the reported experiments show, opening the way for effective applications in retrieval tasks and ontology