Reasoning and Language Generation in the SUMO Ontology
Ramona Enache · Chalmers Publication Library (Chalmers University of Technology) · 2010
We describe the representation of SUMO(Suggested Upper-Merged Ontology)in GF(Grammatical Framework). SUMO is the largest open-source ontology, describing over 10,000 concepts and the relations between them. In addition to this, there are axioms that specify the behaviour of relations and the connections between various concepts. The languages that are widely used for encoding ontologies do not have a type system and have mainly descriptive purpose. For checking the consistency of ontologies or generating natural language, other tools are used. GF is a grammar formalism with support for dependent types, and has built-in support for natural language generation and multilingual translation for 16 languages. The benefits of the translation of SUMO to GF are the possibility to perform type-checking on the content of the ontology, and the generation of syntactically correct natural language. The representation of SUMO uses dependent types for flexibility and better control of semantic actions. The current work provides algorithms for type inference and type checking of the translated axioms. From the concepts, relations and axioms from SUMO, we generate constructions in natural language for English, Romanian and French. The resulting GF files are further more translated to a first-order logic format, TPTP-FOF and checked for consistency with an automated theorem prover. The resulting set of axioms can be used for making inferences. The representation of SUMO in GF preserves the expressivity of the original ontology, adding to this the advantages of a type system and built-in support for natural language generation.