Word Sense Disambiguation for Ontology Learning

Ontology learning aims to automatically extract ontological concepts and relationships from related text repositories and is expected to be more efficient and scalable than manual ontology development. One of the challenging issues associated with ontology learning is word sense disambiguation (WSD). Most WSD research employs resources such as WordNet, text corpora, or a hybrid approach. Motivated by the large volume and richness of user-generated content in social media, this research explores the role of social media in ontology learning. Specifically, our approach exploits social media as a dynamic context rich data source for WSD. This paper presents a method and preliminary evidence for the efficacy of our proposed method for WSD. The research is in progress toward conducting a formal evaluation of the social media based method for WSD, and plans to incorporate the WSD routine into an ontology learning system in the future.