Research And Implementation Of UGC Based Online Classroom And Methods Of Its Resource Storage And Recommendation

Internet classroom is the product of informationization development of educationindustry. The training systems of most domestic enterprises have their onlineclassrooms in PGC based pre-planned course mode. Although this mode is quitemature and easy-to-use, its shortcomings of high cost, long period, monotonouscontent, poor flexibility and slow updating can just not satisfy users’ demands ofricher course content, freer course selection and more flexible time to study. Referringto the ideas of popular UGC based web sites, this dissertation proposed UGC added tothe course mode of training systems, making users served by both PGC and UGC.However, the introduction of UGC shall bring some new problems: high scalability ofsystem architecture required, massive resource data to store and fast accuraterecommendation for users. Redesigning the architecture of the UGC based onlineclassroom with ROA, is a better way to enhance its scalability and portability. And it’sgradually attracting people’s focus to implement mass data storage and real-timeaccurate recommendation using Hadoop for the advantages of low cost and goodperformance.On the premise of finishing a UGC based Internet classroom, with Java as theserver developing language, Flex for stream media application clients and a Hadoopcluster (one master node and two slave nodes) built by Linux virtual machines, taskshas been done as follows:(1) According to the basic concept and features of UGC, the service model wasput forward and in order to fulfill the requirements of multimedia applications andmassive resource storage and recommendation, a runtime architecture model of UGCbased Internet classroom was given.(2) ROA was adopted for system architecture construction, which strictly obeysREST principles, thus the scalability, portability and availability were promotedgreatly.(3) Due to the characteristic “written once and read many times” of massivemultimedia files of courses, the method of storaging them in HDFS was implemented,therefore the problem of mass data storage was solved and the availability wasguaranteed.(4) Users’ visiting records were stored in HBase as the index for user preferenceanalysis, UserCF and ItemCF algorithms were made to run as Map/Reduce jobs in Hadoop cluster, and the detailed Map/Reduce job flow chart and recommendationsystem architecture were given.(5) The methods of developing couples of multimedia applications of Internetclassroom with Red5were elaborated, which provided an open-source solution ofstream media application development.