Research On Leaning Resource Personalized Recommendation System Based On ATCF Algorithm

The e-learning has become a popular way of learning in the information age, it has brought plenty of learning resources, but at the same times, there are a1-ot of problems, the vast of learning resources led to the "information overload" problem, learners need to spend more time to filter the needed resources; the ne-twork learning platform is based on "the goods first" mode which doesn’t consi-der the interest of learners and can’t meet the personalized needs of different le-arners. Personalized recommendation has been applied to e-commerce successfully, so its application to the network teaching, provide personalized service for learn-ers, to solve the problem of "information overload" has become an important res earch content in network teaching.This paper do in-depth introduce on the personalized recommend system and it’s related technology, as to the cold start, data sparseness issue about the colla-borative filtering technology, we summarize the methods existing, and give a co-mbinative commendation based on attribute and time, in order to relieve the pro-blem of cold start, give a arithmetic based on item’properties and user’propert-ies CF, for new items, new users, using its’own properties instead of rating m-atrix to find similar neighbors; in order to relieve the data sparse problem, give a arithmetic of filling the score matrix scheme, through items’attribute and time to fill the score matrix and get a dense ratings matrix, then using users’attrib-utes and scoring to find target user’similar users in the dense ratings matrix, fi nally get the recommendation list according to the similar neighbor’rating; cons-idering the user interest will change with time so joined the forgetting function of psychology to change the scores’weight at different times when calculate the similarity.Implement the proposed combined attribute and time weight recommendation algorithm based on ATCF using the Java programming language, and validate the efficiency using Movielens data. The precision and recall comparison experiments between ATCF and collaborative filtering algorithm shows that the ATCF algorithm is better than the traditional collaborative filtering algorithm.Finally, based on the platform of resources construction with analysis of network learning, the ATCF algorithm is applied to the e-learning system, eliminates the disadvantages existing in the traditional network learning system.