In NUS alone, total undergraduate enrollment increased by 10% to 31,257 in 5 years time from 2015 to 2019. This number is expected to increase further in the coming years. Similar trend is observed across the world. Among all the possible obstacles these students might face, module selection is definitely one of them. Restricted by the limited number of modules that can be taken and motivated to optimize their learning experience, students constantly find it challenging to choose the most suitable courses. There is also a lack of supporting system at the university level, motivating a high demand for module recommenders that can accurately capture students’ preference and recommend suitable modules.
Hence, this project aims to build a module recommender to ease the stress about module selection. The task of module recommendation can be formulated as follows: using module enrollment histories as input, with addition of student and module attributes like student major and module difficulty level, the recommender models the complex student-module interaction pattern and outputs the top 10 modules each student is most likely to take in the future.
Currently, a sample of anonymized student data from NUS is used for building the module recommender. The interim results have been communicated up to the Provost's level.