Filling the last gap: Introducing Multi-Word Expressions to Verb Metaphor Detection

Anonymous

Abstract: Metaphor, as a powerful cognitive modality, possesses the ability to transfer knowledge structures from one domain to another. As metaphor detection continues to receive attention in the field of natural language processing, its importance in downstream tasks such as information extraction, sentiment analysis, and human-computer interaction has gradually become more prominent. However, previous studies have mainly focused on the implicit semantics of individual words, ignoring the fact that combinatorial words may have implicit semantics. In this paper, we propose for the first time a verb metaphor detection task containing multiple words. The goal of this task is to identify verbs or verb phrases with metaphorical usage in a sentence. Subsequently, we introduced a new dataset of verb metaphors. Next, we employed the theory of selection preference violation (SPV) and the metaphor identification program (MIP) for the multi-word verb metaphor task, both of which have been shown to be effective in single-verb metaphor detection. The experimental results show that SPV and MIP can effectively improve the performance of the model on the multi-word verb metaphor detection task.

Paper Type: long

Research Area: Semantics: Sentence-level Semantics, Textual Inference and Other areas

Contribution Types: Data resources

Languages Studied: English