作者: Kaushal Kurapati , Srinivas Gutta
DOI:
关键词: Cluster analysis 、 Third party 、 Value (computer science) 、 Information retrieval 、 Computer science 、 Data set (IBM mainframe) 、 World Wide Web 、 Stereotype (UML)
摘要: A method and apparatus are disclosed for recommending items of interest to a user, such as television program recommendations, before viewing history or purchase the user is available. third party processed generate stereotype profiles that reflect typical patterns selected by representative viewers. can select most relevant stereotype(s) from generated thereby initialize his her profile with closest own interests. clustering routine partitions (the data set) into clusters using k-means algorithm, points (e.g., programs) in one cluster closer mean than any other cluster. The value k incremented until (i) further incrementing does not yield improvement classification accuracy, (ii) predefined performance threshold reached, (iii) an empty detected.