Metaphor Generation with Conceptual Mappings

TL;DR

This paper introduces two novel methods for generating metaphors by leveraging conceptual mappings, improving the quality and meaningfulness of generated metaphoric expressions through lexical and controlled sequence-to-sequence models.

Contribution

It proposes two new approaches, CM-Lex and CM-BART, for metaphor generation based on conceptual metaphor theory, with CM-BART achieving state-of-the-art results.

Findings

CM-BART outperforms existing models in automatic evaluations.

CM-Lex is competitive with recent deep learning systems.

Both methods effectively generate meaningful metaphors.

Abstract

Generating metaphors is a difficult task as it requires understanding nuanced relationships between abstract concepts. In this paper, we aim to generate a metaphoric sentence given a literal expression by replacing relevant verbs. Guided by conceptual metaphor theory, we propose to control the generation process by encoding conceptual mappings between cognitive domains to generate meaningful metaphoric expressions. To achieve this, we develop two methods: 1) using FrameNet-based embeddings to learn mappings between domains and applying them at the lexical level (CM-Lex), and 2) deriving source/target pairs to train a controlled seq-to-seq generation model (CM-BART). We assess our methods through automatic and human evaluation for basic metaphoricity and conceptual metaphor presence. We show that the unsupervised CM-Lex model is competitive with recent deep learning metaphor generation systems, and CM-BART outperforms all other models both in automatic and human evaluations.