Towards adaptation of named entity classification

Numerous state-of-the-art Named Entity Recognition (NER) systems use different classification schemas/ontologies. Comparisons and integration among NER systems, thus, becomes complex. In this paper, we propose a transfer-learning approach where we use supervised learning methods to automatically learn mappings between ontologies of NER systems, where an input probability distribution over a set of entity types defined in a source ontology is mapped to a target distribution over the entity types defined for a target ontology. Experiments conducted with benchmark data show valuable re-classification performance of entity mentions, suggesting our approach as a promising one for domain adaptation of NER systems.