Research Of C Programming Exercises Recommendation Based On Hierarchical Labels

Computer Aided Instruction (CAI) has become one of the powerful and contingent tools as learners’ assistants because of the rapid growth of the Internet. While many researchers have studied various issues in CAI, little work has been done on personalized question recommending for more efficient exercise in online CAI. In this paper, we propose a novel question recommendation system for personalized exercises and apply it to C programming exercises recommendation. The main research work and achievements of this paper include:(1) We propose a method which uses the Natural Language Processing technique to label the C programming questions automatically.(2) We propose a HisaER (Exercise Recommendation based on Historical Answer) model by using hierarchical tag clusters and accomplished the recommendation task by questions re-ranking based on the HisaER model.(3) We propose an algorithm which is called testRank. The testRank algorithm built a graph between the student, tags and exercises based on the students’ historical answer and recommend after weight iteration.(4) We propose a system which is a combination of the online CAI and question recommendation. It can recommend questions for more efficient exercise in online CAI.We have performed several experiments on a data from a real-world Online Aided Teaching System, the results showed our method achieved improvements over the Collaborative Filtering approach.