Network-based recommendation analysis on online judge systems

Online judge (OJ) systems are becoming increasingly popular with the growth of programming competitions and people's enthusiasm for programming. The OJ system provides thousands of programming problems for programming competitions and individual learning. In addition, in the OJ system, learners' needs and preferences for personal information are not expressed, there is no learner scoring problem, and there is little semantic information about problem knowledge. The categories of problems can only be considered by the learner's algorithmic knowledge and experience. Many problems do not have corresponding algorithmic labeling and programming concepts. As an online knowledge base, OJ system has its own characteristics which make it difficult for learners to retrieve relevant resources quickly. Therefore, this paper tries to establish the correlation between problems from the angle of a new social role association, and puts forward a social network based on the co-occurrence link of problems. We associate problems with similar properties to homogenized thought-driven tendencies and explore the underlying patterns of association among different problems. This paper introduces three datasets collected in the actual scene and evaluated them experimentally. Availability: Demo data and codes can be downloaded from https://github.com/wuhayuu/OJ_network_analysis