Constructing a Hybrid Automatic Q&A System Integrating Knowledge Graph and Information Retrieval Technologies

Question-answering (QA) system provides a friendly way for human-computer interaction, which has become an important research direction of smart learning. It provides an easy and individual way for the learner to acquire knowledge. This paper focuses on K-12 education and constructs a hybrid automatic question-answering system which integrates Knowledge Based Question Answering (KB-QA) and Information Retrieval-based Question Answering (IR-QA). The system is built based on Chinese textbooks and a Chinese K-12 knowledge graph (edukg.org). Our QA system covers 9 subjects in K-12 education field, including mathematics, Chinese, geography, history, etc. We evaluate our system on more than 9,000 questions, and achieve average accuracy over 70%. The system could provide effective assistance for teachers’ teaching and students’ learning.