期刊名称：International Journal of Computer Science and Information Technologies
电子版ISSN：0975-9646
出版年度：2012
卷号：3
期号：2
页码：3406-3408
出版社：TechScience Publications
摘要：Recommender system apply various techniques and prediction algorithm to predict user interest on information, items and services from the tremendous amount of available data on the internet. The paper studies various algorithm in weka and the metrices used to evaluate algorithm performance. The basic algorithm or predictive model we use are – simple linear regression, k-nearest neighbours(kNN), naives bayes, support vector machine. We also review the pearson correlation coefficient algorithm and an associative analysis-based heuristic. The algorithms themselves were implemented from abstract class recommender, which was extended from weka distribution classifier class. The abstract class adds prediction method to the classifier. In addition to introducing these techniques we survey their use in recommender system. The paper also analyze the algorithm of user based and item based techniques and some modern recommendation approaches such as context-aware approach, Semantic-based approaches, cross-domain based approaches, peer-to-peer approaches and cross-lingual approaches.
