“Mapping WordNet to the Basic Formal Ontology”, in Buffalo Ontology Research Group Meeting, State University of New York at Buffalo, NY, June 8, 2015.
[ slides ]
We present preliminary work on the mapping of WordNet 3.0 to the Basic Formal Ontology (BFO 2.0). WordNet is a large semantic network linking sets of synonymous words (synsets) by means of semantic relations; it is widely used in natural language processing (NLP) tasks. BFO is a domain-neutral upper-level ontology that represents the types of things that exist in the world and relations between them. BFO serves as an integration hub for more specific ontologies, such as the Ontology for Biomedical Investigations (OBI) and the Cell Line Ontology (CLO). This work aims at creating a lexico‑semantic resource that can be used in NLP tools to perform ontology-related text manipulation tasks. Such tasks include semantic interpretation of natural language texts, word sense disambiguation, and information retrieval. The resource could, for example, be used to find terms in biomedical texts and link them to relevant BFO-based ontologies. Our semi-automatic mapping method consists in using existing mappings between WordNet and an upper-level ontology similar to BFO called KYOTO. The latter allows machines to reason over texts by providing interpretations of the words in ontological terms. Our working hypothesis is that a large portion of WordNet synsets can be semi-automatically mapped to BFO using simple mapping rules from KYOTO to BFO, e.g., ‘accomplishment > process’ and ‘#agentive-social-object > role’. The resulting mappings are to be read as ‘a WN synset X refers to something that is a subtype of BFO type Y’, e.g., the synset ‘immunity.n.02’ refers to a subtype of the BFO type ‘disposition’. We evaluate the method on medical synsets, examine preliminary results, and discuss issues related to the method. We conclude with suggestions for future work.