Leveraging LLM Capabilities for iVC in Game Planning
This research project aims to explore using large language models (LLMs) to develop intelligent virtual characters (iVCs) that enable more natural, contextually appropriate responses and intuitive interactions between users and virtual avatar and create more engaging and compelling user experiences. We will investigate strategies for fine-tuning LLMs on domain-specific data, modeling personality and communication style, and integrating knowledge management to understand the connection between avatars and game players. The approach will involve developing a corpus of text-based interactions between users and virtual characters in training LLMs to generate more contextually relevant responses and finally use machine learning techniques to analyze this data and develop models for the task. Identifying state-of-the-art technologies, such as meta-learning and transfer learning, will also be explored to enhance model performance and efficiency. Our method will involve a combination of research and user studies. Thus, an experimental evaluation and user study to assess the efficiency of the proposed techniques.
CfACTs is a Marie Sklodowska-Curie Action (MSCA) COFUND research centre, co-led by Prof Jian Chang and Prof Jian J Zhang, of the Bournemouth University’s National Centre for Computer Animation (BU NCCA); CfACTs runs from October 2020 to September 2025.
