A Unified Framework for Pun Generation with Humor Principles

TL;DR

This paper introduces a unified framework for pun generation that leverages linguistic attributes like ambiguity, distinctiveness, and surprise, improving the quality of homophonic and homographic puns through a multi-component model.

Contribution

The paper presents a novel integrated approach combining context selection, non-pun language modeling, and pun structure prediction to generate more effective puns.

Findings

Outperforms strong baselines in pun quality

Effective in generating both homophonic and homographic puns

Incorporates linguistic attributes to enhance pun humor

Abstract

We propose a unified framework to generate both homophonic and homographic puns to resolve the split-up in existing works. Specifically, we incorporate three linguistic attributes of puns to the language models: ambiguity, distinctiveness, and surprise. Our framework consists of three parts: 1) a context words/phrases selector to promote the aforementioned attributes, 2) a generation model trained on non-pun sentences to incorporate the context words/phrases into the generation output, and 3) a label predictor that learns the structure of puns which is used to steer the generation model at inference time. Evaluation results on both pun types demonstrate the efficacy of our model over strong baselines.