Intelligent Security Q&A System Based on Large Language Model
Youtao Zhou, Qiuhong Lu, Haoyu Fan, Yuntao Xiao, Jinwen Hu, Shimian Zhang · 2024
With the continuous development of artificial intelligence technology, intelligent Q&A systems have been widely used in various fields, including hotels, airports, hospitals and other places. However, traditional Q&A systems are often constrained by the limited rule base and stereotypical dialog flow, and are unable to provide a sufficiently intelligent interaction experience. The system in this paper utilizes Large Language Model (LLM) combined with face recognition technology to build a more intelligent and flexible security Q&A system. By combining the local knowledge base with LLM, the system is able to retrieve relevant information from the local knowledge base based on the professional questions provided by the user through text vectorization and then similarity comparison, and generate corresponding responses, thus providing a more humanized and accurate service. The system has the following capabilities: (1) specialized Q&A, based on LLM and self-constructed knowledge base to generate Q&A with domain-specific knowledge, compared to model fine-tuning methods, this method can deploy large vertical domain models without re-training;(2) Security welcome, this system face recognition module adopts Python+Dlib method to identify the user and conduct the welcome greeting, if the visitor is an unfamiliar user, it is not possible to conduct the Q&A with the LLM for the sensitive information, e.g., private information, etc.; if the visitor is a known user, then it is possible to communicate with the LLM in all aspects.