Online Programming Education Modeling and Knowledge Tracing

With the development of computer technology, more and more people begin to learn programming. And there are a lot of platforms for programmers to practice. It's often difficult for these platforms to customize the needs of users at different levels. In this paper, we address the above limitations and propose an intelligent tutoring model, to help programming platforms achieve better tutoring for different levels of users. We first devise a novel framework for programming education tutoring which is combined with programming education knowledge graph, crowdsourcing system and online knowledge tracing. Then, by ontology definition, information extraction and data fusion, we construct a knowledge graph to store the data in a more structured way. During the knowledge tracing stage, we extract behavior features and question knowledge features from a relational database and knowledge graph separately. Meanwhile, we improve the process for student ability evaluation and adapt the Knowledge Tracing algorithm to predict students' behavior on knowledge and questions. Experiment results on real-world user behavior data sets show that through the help of Knowledge Tracing algorithm, we can achieve considerably satisfied results on students' behavior prediction.