With the development of generative models, research on human-AI co-creation has been actively conducted. However, in the field of co-creation, research on system personalization according to individual characteristics is insufficient, and little focus has been placed on individual differences in creation. Therefore, in this study, we constructed StoryCCDial, a co-creation dialogue dataset aimed at the personalization of co-creative dialogue systems. First, we collected human-human story co-creation dialogue data involving 120 workers and constructed a dataset that includes dialogues, dialogue acts, the workers’ personality traits, postsurveys, and edit histories from the interface. Next, using the constructed dataset, we conducted analyses focusing on the workers’ personality traits, the number of utterances, and edit histories. The analysis revealed differences in dialogue content based on workers’ personality traits, individual differences in the number of utterances during the co-creation process, and variations in creative workflows on the interface. Our dataset will be available at https://github.com/UEC-InabaLab/StoryCCDial .

The Werewolf Game is a communication game where players’ reasoning and discussion skills are essential. In this study, we present a Werewolf AI agent developed for the AIWolfDial 2024 shared task, co-hosted with the 17th INLG. In recent years, large language models like ChatGPT have garnered attention for their exceptional response generation and reasoning capabilities. We thus develop the LLM-based agents for the Werewolf Game. This study aims to enhance the consistency of the agent’s utterances by utilizing dialogue summaries generated by LLMs and manually designed personas and utterance examples. By analyzing self-match game logs, we demonstrate that the agent’s utterances are contextually consistent and that the character, including tone, is maintained throughout the game.