A novel pattern matching method for Chinese metaphor identification and classification

Abstract

Metaphor is a pervasive phenomenon in natural language. This paper focuses on metaphor identification. Based on the linguistic definition, an improved classification system of Chinese metaphors is presented, classifying metaphorical language into two categories: referential metaphor with ten subclasses and collocational metaphor with five subclasses. Furthermore, a novel pattern matching algorithm on dependency tree is proposed for detecting potential metaphor occurrences and two quantified measures, referential adaptability degree (RA) and colloactional adaptability degree (CA), are introduced for determining metaphor category. Finally, a computational model for Chinese metaphor identification and classification is developed. © 2011 Springer-Verlag.