User Centred Ontology Learning

Automatic (or semi-automatic) ontology building is a current issue in many application fields where ontologies are currently built manually. This paper presents a user-centred methodology for ontology construction based on the use of Machine Learning and Natural Language Processing. In our approach, the user starts the process by initially sketching a preliminary ontology (or selecting an existing one) and a corpus of relevant texts. Then the learner uses the ontology in order to retrieve examples of lexicalisation of relations (e.g. ISA relation) in the corpus. Retrieved examples are validated by the user and used by the learner to generate patterns to discover other instances of the same relation. New instances added to the existing ontology or used to tune the existing ontology. The discovering process is repeated until a satisfying ontology is obtained. The methodology largely automate the building process. It focuses the expensive user activity on sketching the initial ontology, validating textual examples and the final ontology, while the system performs the tedious and expensive activity of searching a large corpus for knowledge discovery. Moreover the output of the process is not only an ontology, but also a system trained to rebuild and eventually retune the ontology, as the learner is adapted by the user feedback. This simplifies ontology maintenance, a major problem in ontology-based methodologies.