Parallel user profiling based on folksonomy for Large Scaled Recommender Systems : an implimentation of Cascading MapReduce

摘要

The Large scaled emerging user created informationudin web 2.0 such as tags, reviews, comments and blogs can beudused to profile users’ interests and preferences to makeudpersonalized recommendations. To solve the scalabilityudproblem of the current user profiling and recommenderudsystems, this paper proposes a parallel user profiling approachudand a scalable recommender system. The current advancedudcloud computing techniques including Hadoop, MapReduceudand Cascading are employed to implement the proposedudapproaches. The experiments were conducted on Amazon EC2udElastic MapReduce and S3 with a real world large scaleduddataset from Del.icio.us website.