HumorPlanSearch: Structured Planning and HuCoT for Contextual AI Humor

2025-08-15

AI Summary

Current large language models (LLMs) often generate humor that is generic, repetitive, or culturally inappropriate due to insufficient modeling of contextual dependencies. To address this, we propose HumorPlanSearch—a novel framework that explicitly models context throughout the entire humor generation pipeline. It introduces Humor Chain-of-Thought (HuCoT), a structured planning template; integrates knowledge graph–based strategy reuse, semantic novelty filtering, and judge-driven iterative refinement; and employs multi-signal evaluation fusion for robust assessment. Experiments across nine diverse topics demonstrate that the full pipeline achieves a statistically significant 15.4% average improvement in Humor Generation Score (HGS) over strong baselines (p < 0.05). This yields substantial gains in contextual appropriateness, discourse coherence, and personalization. Our work establishes a new paradigm for culturally sensitive, interpretable, and context-aware AI-driven comedy generation.

Technology Category

• Natural Language Processing: Generation
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Automated humor generation with Large Language Models (LLMs) often yields jokes that feel generic, repetitive, or tone-deaf because humor is deeply situated and hinges on the listener's cultural background, mindset, and immediate context. We introduce HumorPlanSearch, a modular pipeline that explicitly models context through: (1) Plan-Search for diverse, topic-tailored strategies; (2) Humor Chain-of-Thought (HuCoT) templates capturing cultural and stylistic reasoning; (3) a Knowledge Graph to retrieve and adapt high-performing historical strategies; (4) novelty filtering via semantic embeddings; and (5) an iterative judge-driven revision loop. To evaluate context sensitivity and comedic quality, we propose the Humor Generation Score (HGS), which fuses direct ratings, multi-persona feedback, pairwise win-rates, and topic relevance. In experiments across nine topics with feedback from 13 human judges, our full pipeline (KG + Revision) boosts mean HGS by 15.4 percent (p < 0.05) over a strong baseline. By foregrounding context at every stage from strategy planning to multi-signal evaluation, HumorPlanSearch advances AI-driven humor toward more coherent, adaptive, and culturally attuned comedy.

Problem

• Addresses generic humor in LLMs by modeling cultural context
• Improves joke relevance through structured planning and filtering
• Enhances comedic quality with multi-signal evaluation metrics

Innovation

• Plan-Search for topic-tailored humor strategies
• Humor Chain-of-Thought templates for cultural reasoning
• Knowledge Graph retrieval of high-performing historical strategies