Educational Psychology Question Answering System Based on BERT Pre-training Language Model

In this paper, we investigate an intelligent question-answering system based on knowledge graphs.We first discuss the construction of knowledge graphs, the entity recognition scheme, and a novel question parsing algorithm models.Specifically, we employ the BiLSTM-CRF model for named entity recognition, which integrates Bidirectional Long Short-Term Memory Networks (BiLSTM) and Conditional Random Fields (CRF) to significantly enhance recognition accuracy and efficiently capture the contextual information.For question parsing, We use the BERT pre-trained language model, which can achieve an exact matching between questions and templates based on the high-dimensional vector encoding.In order to enhance the system performance and user experience, we introduce the Model-View-Controller (MVC) scheme to optimize our proposed system architecture.In particular, the data layer is responsible for data access, the logic processing layer handles algorithm and logic processing, and the user interaction layer provides a friendly interface and interaction method.The experiment results show the proposed intelligent question-answering system has made significant progress in entity recognition and question parsing and can achieve efficient operation and user-friendly interaction experience.