Research on Large Language Models for Elderly Travel Education Integrated with External Knowledge Bases

A Large Language Model has demonstrated exceptional performance across a variety of tasks. However, when deployed in specific domains like smart elderly care, the models still face challenges due to a deficiency in domain-specific knowledge and encounter performance bottlenecks in providing services for older people. In this paper, we introduce an LLM named ChatDriver, which is tailored for elderly travel education. Considering the importance of data quality, we have carefully designed a fine-tuning dataset for the older population care domain. Additionally, we propose a method that combines vector database retrieval with keyword retrieval to effectively address the uncertainty issues caused by vector similarity matching in vector databases. Furthermore, by employing a novel attention mechanism and improving memory management and computational resource utilization, we have significantly enhanced the model's performance.