The Research And Application Of MOOC Hybird Recommendation Algorithm Based On Big Data Platform

Thanks to the rapid development of the Internet,tremendous changes have been taking place in the traditional field of education.In recent years,MOOC is becoming increasingly popular,which represents a brand new educational method that features no threshold,low cost and abundant learning resources.However,the rapid development of MOOC platform facilitates the substantial increase of MOOC courses,brings about information overload.Consequently,"elective difficulties" occurs: Users find it hard to select the courses they need from a mass of MOOC courses.Therefore,the intelligent algorithm is an essential solution to the information overload of the MOOC platform as it can help the user to select the appropriate courses and highlight the outstanding courses.Recommendation system is considered to as a more efficient approach to solve information overload.Although the recommendation system has been successfully applied in many areas,few researches have been made on the system application in the field of MOOC at home and abroad.The accuracy of course recommendation results is bound to stay at a relatively low level if we mechanically copy previous usage experience and do not consider the scene characteristics of MOOC applications.To solve "elective difficulties" of MOOC platform,this paper proposes an implicit scoring model for MOOC,and designs to release a MOOC recommendation system based on big data platform for the current Internet big-data environment.The main contributions and innovations are listed as follows:(1)Propose an implicit scoring model for MOOC.The model considers the application scenario characteristics of MOOC platform: We've gained experience from the user learning behavior and successful application of other recommendation systems in the past.(2)Use the implicit scoring model for MOOC to optimize the traditional itembased collaborative filtering recommendation algorithm and matrix decomposition algorithm.The experimental results show that the implicit scoring model for MOOC can improve the recommendation accuracy of those two traditional recommendation algorithms in MOOC applications.(3)Design a MOOC recommendation system based on big data platform to deal with the big data environment of Internet.We divide the whole system into six modules according to the business characteristics of big-data MOOC application.Each module adopts micro-service architecture,enabling more convenience for further system extension and maintenance.(4)Provide the parallelization solution of the collaborative filtering algorithm based on the implicit scoring model for MOOC by taking advantage of the MapReduce computing model.Then,we use Spark MLlib to implement the matrix decomposition algorithm based upon the features of the iterative algorithm,which greatly reduces the computing time and enhances processing capacity against large-scale data sets.