English online education resource recommendation using knowledge graph and ARA algorithm

Abstract

This paper is the first time to build English education resource network, which significantly improves the accuracy and personalization of resource recommendation on English online education platform. The main contribution of this paper is the successful construction of English education resource network using knowledge graph, and the adaptive recommendation algorithm (ARA) to provide users with customized learning resources based on the in-depth analysis of user behavior data and course content. In terms of data collection and processing, we have integrated more than 100,000 pieces of user learning behavior data, including users’ learning history, preferences, and resource interaction. Based on these data, we use knowledge graph technology to semantically represent educational resources and build an information network containing rich associations. This network not only covers the logical relationship between knowledge points but also includes the interaction relationship between users and resources, which provides a solid foundation for subsequent recommendation algorithms. In the aspect of recommendation algorithm, we adopt adaptive recommendation algorithm. The algorithm can filter out the learning resources that best meet the user’s individual needs from the knowledge graph according to the user’s learning progress, historical behavior, and current needs. Through comparative experiments with other recommendation algorithms, we found that the ARA algorithm improves the accuracy rate by 25% and the recall rate by 18%, which is significantly better than the traditional recommendation method. Specifically, when users access the online education platform, the system will quickly retrieve relevant learning resources from the knowledge graph based on their historical learning data and current needs, and sort and recommend them based on the ARA algorithm. The experimental results show that the system can recommend more than 80% of the content that users are really interested in, and cover more than 90% of user needs, which significantly improves the user’s learning experience and efficiency. This study not only provides a new resource recommendation method for online English education platforms but also provides a valuable reference for personalized recommendation research in other fields.