LOLGORITHM: Funny Comment Generation Agent For Short Videos

2026-04-09

AI Summary

Existing approaches struggle to generate authentic comments that align with the cultural norms, linguistic conventions, and high-engagement expectations of short-video platforms. To address this challenge, this work proposes a modular multi-agent framework that integrates video content understanding, semantic retrieval, and a trending-meme enhancement mechanism to support controllable generation across six distinct stylistic dimensions. The study also introduces the first bilingual, high-engagement comment dataset specifically curated for short-video platforms and designs a decoupled, generalizable architecture enabling seamless collaboration among multiple functional modules. Experimental results demonstrate that the generated comments achieve human preference rates of 80.46% on YouTube and 84.29% on TikTok (Douyin), significantly outperforming baseline methods, with performance robust across different large language models.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation
• Computer Vision: Multi-modal Vision

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search

Abstract

Short-form video platforms have become central to multimedia information dissemination, where comments play a critical role in driving engagement, propagation, and algorithmic feedback. However, existing approaches -- including video summarization and live-streaming danmaku generation -- fail to produce authentic comments that conform to platform-specific cultural and linguistic norms. In this paper, we present LOLGORITHM, a novel modular multi-agent framework for stylized short-form video comment generation. LOLGORITHM supports six controllable comment styles and comprises three core modules: video content summarization, video classification, and comment generation with semantic retrieval and hot meme augmentation. We further construct a bilingual dataset of 3,267 videos and 16,335 comments spanning five high-engagement categories across YouTube and Douyin. Evaluation combining automatic scoring and large-scale human preference analysis demonstrates that LOLGORITHM consistently outperforms baseline methods, achieving human preference selection rates of 80.46\% on YouTube and 84.29\% on Douyin across 107 respondents. Ablation studies confirm that these gains are attributable to the framework architecture rather than the choice of backbone LLM, underscoring the robustness and generalizability of our approach.

Problem

• short-form video
• comment generation
• platform-specific norms
• user engagement
• cultural and linguistic authenticity

Innovation

• multi-agent framework
• stylized comment generation
• meme augmentation
• short-form video
• human preference evaluation