The Knowledge Representation for Comparison Words in Extended-HowNet

In this paper, we study the semantic representations of comparison words and comparative constructions under the frameworks of E-HowNet and to see how the mechanism of semantic composition works. Our methodology is to establish a mapping between grammatical structures and fine-grained event structures for comparative constructions. We encode event structures of comparative constructions as parts of representations of comparison words. The fine-grained semantic roles are adopted from FrameNet. A semantic composition mechanism was established to unify word sense representations under syntactic constraints. Coercion and filling semantic gaps are integral parts of the process. Fully automatic semantic composition will be our ultimate future goal.