Mining Unexpected Sequential Patterns from MOOC Data
Wei Guo Song, Wei Ye · 2021
Massive Open Online Courses (MOOCs) are triggering a revolution in education, and the learning data accumulated by MOOC platforms contain a huge wealth of information for improving online learning effects. To improve learners' experience, we propose an algorithm for mining unexpected sequential patterns. First, we design unexpected support to replace support. Unexpected support also satisfies the downward closure property and ensures that each input sequence does not contribute equally to the measurement of sequential patterns. Second, we introduce an item list and sequence list according to the characteristics of MOOC data. The lists not only speed up the generation of new candidates but are also used for narrowing the search space. Third, we describe and explain the proposed algorithm FAST-USP in detail. Finally, we report the experimental results on efficiency, memory consumption, and the number of results, which demonstrate the superiority of the FAST-USP algorithm.