Exploration and Practice of Applying Large Language Models in Home Education Guidance

Currently, large language models (LLMs) have demonstrated remarkable capabilities in general question answering, text generation, and scientific reasoning, profoundly influencing the development of natural language question answering. In this context, this study proposes a new model that combines a large language model with family education guidance, and constructs a personalized guidance question-and-answer system for family education. To enhance the professionalism and accuracy of the model in family education guidance question answering, we autonomously built a knowledge base for family education guidance and a dataset for family education guidance dialogues. Simultaneously, a prefix text filtering network was introduced, coupled with memory storage techniques, effectively preventing hallucination issues. Finally, experiments were conducted through expert evaluation and performance assessment, confirming the system's professionalism and effectiveness in family education guidance. In comparison to general language models like ChatGPT, this system has achieved significant results in providing guidance on family education and answering questions related to professional family education knowledge.