Long-Term Memory for Large Language Models Through Topic-Based Vector Database

Large language models (LLMs) have garnered sub-stantial attention and significantly transformed the landscape of artificial intelligence, due to their human-like understanding and generation capabilities. However, despite their excellent capabilities, LLMs lack the latest information and are constrained by limited context memory, which limits their effectiveness in many real-time applications that require up-to-date information, such as personal AI assistants. Inspired by the recent study on enhancing LLMs with infinite external memory using vector database, this paper proposes a topic-based vector database to enable LLMs to achieve long-term personalized memory. By leveraging prompt engineering to fully utilize the semantic understanding capabilities of LLMs, an efficient topic-based per-sonalized memory management system is designed to store and update user's preferences and characteristics. This system can be applied in various AI assistant domains, such as companion robots, to efficiently store personal memories of users through conversations, ultimately fulfilling their needs in a personalized manner.