Design of a Chinese Learning Platform Based on Knowledge Graph and Adaptive Question Answering

Abstract

To address fragmented knowledge structures, unclear learning paths, and insufficient personalized support in Chinese learning for non-native speakers, this paper designs a Chinese learning platform integrating knowledge graph technology and adaptive question answering. First, a multidimensional Chinese knowledge graph covering vocabulary, grammar, and culture is constructed, and semantic links among knowledge nodes are established through entity-relation extraction. Second, a learner ability modeling module based on Item Response Theory is developed to dynamically evaluate learners’ proficiency levels. An adaptive question-answering mechanism then recommends exercises with appropriate difficulty according to learners’ current ability and knowledge dependencies in the graph. Finally, a reinforcement learning algorithm is introduced to optimize the recommendation strategy. Experimental results show that learners using the platform achieve an increase in learning efficiency of 18.5 questions per hour and a 12.6 percentage-point improvement in knowledge mastery accuracy. User satisfaction with personalized recommendation reaches 89.2%, with a mean score of 4.35, indicating the effectiveness of the proposed intelligent learning platform.