Story Generation in Multi-Agent Systems: A DynamicCollaboration Framework Based on Reinforcement Learning
DOI:
https://doi.org/10.54097/4m16y426Keywords:
Multi-Agent Systems; Story Generation; Reinforcement Learning; Transformer, Narrative AI.Abstract
In artificial intelligence, multi-agent systems excel in complex tasks requiring coordination. This paper presents a Multi-Agent Dynamic Collaboration Framework for story generation, tackling narrative coherence, diversity, and adaptability. Using GPT-2 agents for plot, dialogue, and description, a Transformer-based coordinator dynamically adjusts contributions via multi-head attention. PPO reinforcement learning optimizes behaviors with rewards based on BLEU, ROUGE, and entropy. Trained on ROCStories, the framework achieves BLEU 0.85, ROUGE-L 0.65, and perplexity 12.5, surpassing baselines. Human assessments show improved coherence and engagement. This advances AI creative writing and enables scalable applications in education, entertainment, and marketing for immersive narratives.
Downloads
References
[1] J. P rez et al., “Serious games and ai: Challenges and opportunities for compu- tational social science,” IEEE Access, vol. 11, p. 10153609, 2023. DOI: https://doi.org/10.1109/ACCESS.2023.3286695
[2] S. Joksimovic et al., “Opportunities of artificial intelligenc for supporting complex problem-solving: Findings from a scoping review,” Computers and Education: Artificial Intelligence, vol. 4, p. 100138, 2023. DOI: https://doi.org/10.1016/j.caeai.2023.100138
[3] Y. Liu et al., “Benchmarking large language model-based agent systems in healthcare,” npj Digital Medicine, vol. 9, pp. 1–10, 2026. DOI: https://doi.org/10.1038/s41746-026-02443-6
[4] Y. Xie et al., “Rapid generation method of process routes based on multi-agent collaboration with llms,” Advanced Engineering Informatics, vol. 62, pp. 101– 115, 2025. DOI: https://doi.org/10.1016/j.aei.2025.103733
[5] X. Chen et al., “Collaborative causal inference and multi-agent dynamic intervention in dual-carbon public opinion analysis,” Systems, vol. 13, no. 8, p. 689, 2025. DOI: https://doi.org/10.3390/systems13080689
[6] R. Łabedzki, “Human-ai collaboration in hybrid multi-agent systems,” International Journal of Electronics and Telecommunications, vol. 71, no. 1, pp. 1–9, 2025. DOI: https://doi.org/10.24425/ijet.2025.155474
[7] F. Huot et al., “Agents’ room: Narrative generation through multi-step collabo- ration,” Proceedings of the AAAI Conference on Artificial Intelligence, vol. 39, pp. 18 450–18 458, 2025.
[8] Y. Che et al., “Steering narrative agents through a dynamic cognitive framework for guided emergent storytelling,” Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, vol. 21, pp. 1–10, 2025. DOI: https://doi.org/10.1609/aiide.v21i1.36841
[9] X. Liang et al., “Cmat: A multi-agent collaboration tuning framework for enhancing small language models,” Proceedings of the International Conference on Learning Representations, pp. 1–15, 2025.
[10] S. Y. Shengbin et al., “Multi-agent simulator drives language models for legal reasoning,” Findings of the Association for Computational Linguistics: NAACL 2025, pp. 365–380, 2025.
[11] W. Epperson et al., “Interactive debugging and steering of multi-agent ai systems,” Proceedings of the ACM on Human-Computer Interaction, vol. 9, pp. 1–25, 2025. DOI: https://doi.org/10.1145/3706598.3713581
[12] G. Nettem et al., “Agentflow: A context aware multi-agent framework for dynamic agent collaboration,” Proceedings of the International Conference on Intelligent Systems and Applications, p. 133757, 2025. DOI: https://doi.org/10.5220/0013375700003890
[13] C. Li et al., “Development of ai multi-agent frameworks for financial services: A refined perspective,” Journal of Innovation in Engineering and Applied Sciences, vol. 3, no. 6, pp. 1–15, 2025. DOI: https://doi.org/10.70393/6a69656173.333337
[14] H. Yin et al., “Dynamic and multi-role contexts in recommendation systems,” IEEE Transactions on Knowledge and Data Engineering, vol. 38, pp. 1–14, 2026.
[15] P. Sadler, “Toward learning collaborative visual referring: bootstrapping neural architectures,” Neural Networks, vol. 170, pp. 1–20, 2025.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Highlights in Science, Engineering and Technology
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
