Resumen de µ-Ontologies: integration of frame semantics and ontological semantics
Guntis Barzdins, Normunds Gruzitis, Gunta Nespore, Baiba Saulite, Ilze Auzina, Kristine Levane-Petrova
Today FrameNeta-a state-of-the-art implementation of frame semantics-provides one of the best insights into lexical semantics and their interaction with the syntactic structure of the sentence.
The main limitation of the current implementation is the insufficient level of formalization of frame descriptions, making it unsuitable for automatic text annotation without human supervision.
Meanwhile, FrameNet usability would greatly benefit from more rigorous formalization and the consequential possibility for automatic annotation. Previous attempts at formalization have focused on enforcing strict ontological control of the semantic types for the frame fillers-despite their insignificant use-due to high ambiguity-in the actual FrameNet. We propose a different approach relying on representation of FrameNet as a 4D multidimensional ontology that allows capturing of the precedent. knowledge encoded in the manually annotated texts, like FrameNet's full-text annotation reports. This allows both to re-create FrameNet ontology from semantically annotated texts, as well as to use this representation for semantic annotation of new texts. Further extensions of this approach with 5th dimension for anaphora annotation is discussed as an alternative for the informal semantic type mechanism of FrameNet.
