Recognition and Extraction of Intentions Based on Ontology

This paper describes a system that allows recognizing author´s intentions from electronic scientific publications. It is based on a semi-automatic analyzer that uses an ontology of intentions and applied on a mono-domain textual corpus. The recognition mechanism is based on a set of algorithms that facilitate the building of intentional structure i.e. the segmentation of a document according to the author´s intentions. A knowledge base that describes intentions in a specific domain is used by the analyzer to facilitate the recognition. It is input in the system during the initialization phase by a domain expert. The system identifies verbs and their intentional concept, and then updates the knowledge base of intentions for its enrichment. The paper presents also an experiment on scientific publications in the field of computer sciences.