Design and Implementation of Writing Recommendation System Based on Hybrid Recommendation

The personalized recommendation model is widely applied and has achieved good results in many areas. However, considerable effort is still necessary in education-related recommendations. Therefore, based on a website platform for TOEFL and IELTS writing and some traditional recommendation algorithms, this paper designs an algorithm of hybrid recommendation by considering users’ scores and background information, which handles the cold start problem as well. Specifically, the matrix of user and item is filled with users’ writing scores, and the default value is replaced by the reliable average scores announced by the authorities of TOEFL and IELTS. Such scores are a solution to data sparsity in a certain degree. Then, personalized writing topics are recommended to new users by calculating the similarity of background information between the new and old users. Experimental results show that the proposed scheme has better recommendations compared with other existing algorithms. With these advantages, we design a web-based hybrid recommendation system for online writing with multiple functions.