Design And Implementation Of Movie Recommendation System Based On Partition Mixed Mode

With the vigorous development of the Internet,data overload is becoming more and more serious,so it is of great value to mine information quickly and efficiently.The recommendation system emerging in recent years combines information mining and artificial intelligence technology,which meets the needs of network information mining.In the traditional recommendation system,there are some problems such as cold start,data sparsity and scalability,which restrict the further development of technology.At the same time,the single recommendation algorithm is ineffective and can no longer meet the growing needs of users.In view of the above problems,after investigating a large number of recommendation systems and big data processing frameworks,this paper proposes a movie recommendation system based on partition mixed mode,which is implemented on Spark distributed platform.The related research work is as follows:(1)To solve the problem of sparse data of LFM algorithm on small sample data sets,UGC algorithm based on TF-IDF is introduced for optimization,and UGC and LFM are fused as a new offline recommendation algorithm.Experimental results show that compared with the single LFM algorithm,the three performance evaluation indexes of the improved algorithm are improved to some extent.Therefore,the defects of single algorithm can be compensated by algorithm fusion,and UGC algorithm plays an auxiliary role in correcting LFM algorithm errors.(2)Aiming at the defects of the mainstream cosine similarity online recommendation algorithm in Spark ecosystem,geodesic measurement in manifold space is introduced and the corresponding algorithm is implemented.The proposed new recommendation algorithm not only considers the difference of numerical values,but also takes into account the spatial structure of movie data distribution.Experimental results show that,as a beneficial combination of information geometry and online recommendation of big data,the proposed geodesic algorithm has obvious advantages in three performance indexes compared with traditional cosine similarity algorithms.(3)Based on the above improved offline and online recommendation algorithm,a complete and user-friendly movie recommendation system is designed and built in strict accordance with the thinking of software engineering,from the two aspects of implementation and performance.The whole recommendation system is designed based on the partition mixed mode,and consists of online recommendation area,offline recommendation area and popular statistical recommendation area.It has common basic functions such as view,score and search.The system can stably and efficiently complete the preset recommendation,which verifies the feasibility and effectiveness of the partition hybrid mode application.