A Unified Framework for Pun Generation with Humor Principles
This work addresses the challenge of automated pun generation for natural language processing applications, though it appears incremental as it builds on existing methods for pun types.
The authors tackled the problem of generating both homophonic and homographic puns by proposing a unified framework that incorporates linguistic attributes like ambiguity, distinctiveness, and surprise into language models. Their model outperformed strong baselines in evaluations on both pun types.
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.