The retrieval system is a crucial module in e-commerce search advertising that matches user queries with ads. The diverse expressions of users often produce massive tail queries that cannot match merchant bidwords, leading to poor retrieval efficiency. Existing methods, such as query log mining and vector matching, fail to optimize relevance, authenticity, and ad revenue of the rewrite.

With the help of Internet, Massive Open Online Courses (MOOC) are recognized as a new path to learn courses via the web instead of in the traditional classrooms. MOOC can break many limits such as distance, time, participants, on the traditional courses. At the same time, it brings some new issues, such as high drop out ratio. Nowadays increasing MOOC courses are available and even more common people are involved into this kind of new learning procedure. How to evaluate the learning behaviors of MOOC is still an open problem. We propose an efficient algorithm to cluster the MOOC learning events into many closely related sets and name such set as LES (Learning Events Set) to model one basic learning procedure on MOOC. The quality of LES is highly dependent on the maximum time period Tmax between two LESes. We systematically investigate this problem and propose an efficient method to set the value of Tmax. Our method has been employed into one MOOC platform, XuetangX and the experimental results demonstrate that our method can really work.