文章基本信息

标题：Cuckoo Search Based Personalized View for Movie Recommendation over Social Networks
本地全文：下载
作者：S. Uma Shankari ; M. Chidambaram
期刊名称：Research Journal of Applied Sciences
印刷版ISSN：1815-932X
电子版ISSN：1993-6079
出版年度：2020
卷号：15
期号：3
页码：102-111
DOI：10.36478/rjasci.2020.102.111
语种：English
出版社：Medwell Journals
摘要：Owing to the exponential growth ofinformation in online social networks, the users of suchnetworks demand the recommendation systems to deliversignificant results. A recommendation system rightlysuggests the personalized movies that are desirable to theusers predominantly from large information storage.Notably, the current research works in movierecommendation system focus on determining the mostrelevant features from the user profile information andshared contents in the social network. Even though theexisting research works recommend the movies that are inproximity to the user preferences, there is a profound needfor further exploring the features of the movie and thusensure the highly desired movies to the users. Hence, thispaper targets on recommending the movies with theknowledge of analyzing the movie features along with thedata clustering and computational intelligence methods.This article proposes the Cuckoo search based MOstpersonalized VIEw in item recommendation (CMOVIE)model, incorporating the missing ratingprediction and contextual movie recommendation phases.At first, the C-MOVIE approach explores the features ofthe movies to recognize the interest of the users in termsof inherent features after reducing the featuredimensionality by Principal Component Analysis (PCA)method. Then, it clusters the users based on therecognized features by K-means clustering and Cuckoosearch optimization methods with the intention ofgrouping the users with similar interests which eases themissing rating prediction when using Probabilistic MatrixFactorization (PMF). In the end, the C-MOVIE approachcontextually recommends the movies to the users bymapping the features of the new movies with the featuresof the clustered users. The experimental results yieldedon Douban movie which data set demonstrate that the CMOVIEapproach distinctively delivers the personalizedmovie recommendation than the existing HPSO method.
关键词：Recommender system;movie;features;clustering;context and probabilistic matrix factorization