Research On Personalized Recommendation Method Of Live Broadcast Platform Based On RFM Model

With the development of the live broadcast industry becoming more stable,major live broadcast platforms have begun to focus on refined operations,and strive to further improve user loyalty while providing quality content for the live broadcast platform.Among them,paying users are the key targets for live broadcast platforms.Improving user loyalty,one is that the live broadcast platform continuously provides high-quality live content,and the other is to recommend suitable live content for users according to user preferences,reducing the cost of user interest discovery.This article takes the live broadcast platform as the research object.Incorporate customer relationship management into the recommendation method of the live broadcast platform.Based on this,a customer value evaluation model is introduced,and the user is grouped according to user value.Recommended to create higher profits for the enterprise while increasing user loyalty.First,the article adopts the RFM model as the basis of user value grouping,and incorporates the user behavior data unique to the live broadcast platform-the average viewing time,increases consideration of user stickiness and long-term value,thus forms the RFMT model.Based on the RFMT model,RFMTCluster and RFMTSegmentation are proposed to segment users of different values.Then,this paper adopts the index weighting method to calculate the user-anchor value preferences,and then constructs the user's rating matrix,and uses the collaborative filtering algorithm to conduct recommendation experiments.Finally,through three sets of experimental comparisons,the following results are obtained: 1)The user segmentation method based on RFMTSegmentation is better than the kmeans clustering algorithm More suitable for practical application recommendations.2)The recommendation effect of various groups after subdivision has been greatly improved than before the subdivision.The F1 values of core users,important users,general users and potential users have increased by 2.73%,1%,0.4%,0.72%.