Explainable Metaphor Identification Inspired by Conceptual Metaphor Theory

Metaphor is not only a linguistic phenomenon but also reflects the concept projection between source and target domains in human cognition. Previous sequence tagging-based metaphor identification methods could model projection, resulting limitation that outputs of these models are unexplainable predictions metaphoricity labels. In this work, we propose first explainable model, inspired by Conceptual Theory. The based on statistic learning, lexical resource, novel reward mechanism. Our can identify word-pair level, explain predicted labels via learned mappings. use mechanism allows to learn optimal mappings without knowing their true method applicable for concepts out training using resource. automatically generated demonstrate implicit thoughts metaphoric expressions. experiments show effectiveness proposed identification, mapping tasks, respectively.