马学明 童怀



摘要:近年来,各类视频应用上内容越来越丰富,页面上与当前用户无关的内容也越来越多。因此,市面上出现了多种不同的推荐算法来进行内容推荐。但是,不是每种推荐算法都能够解决所有的问题。基于个性化推荐系统的视频App,融合了多种推荐方法。首先为了解决推荐系统的冷启动问题,采用了基于统计学的推荐方式,同时,采用基于协同过滤的推荐算法,计算视频和用户间的隐藏特征,最后还有实时推荐模块,能够根据用户近期的行为对推荐内容进行调整。
关键词:android应用;推荐系统;协同過滤;ALS算法;实时推荐
中图分类号:TP391 文献标识码:A
文章编号:1009-3044(2021)08-0004-03
Abstract: In recent years, the content of various video applications is becoming more and more abundant, and there are more and more content on the page that has nothing to do with the current users. Therefore, there are many different recommendation algorithms in the market for content recommendation. However, only use a recommendation algorithm can not solve all the problems. The video App based on personalized recommendation system integrates various recommendation methods. Firstly, in order to solve the cold start problem of the recommendation system, a recommendation method based on statistics is adopted. At the same time, a recommendation algorithm based on collaborative filtering is adopted to calculate the hidden features between video and users. Finally, there is a real-time recommendation module, which can adjust the recommended content according to the recent behavior of users.
Key words: android application; aecommendation system; aollaborative filtering; ALS algorithm; real-time recommendation
1 背景
近年来,为了满足人们越来越高的生活需求,各类视频App层出不穷,有爱奇艺视频、腾讯视频等长视频应用,也有抖音、快手之类的短视频应用。这些应用上各种内容非常丰富,但同时也充斥着大量与用户无关的内容。因此,为了能够在大量的网络资源中找到用户最感兴趣的内容,个性化推荐系统也就应运而生。
视频推荐系统常采用基于协同过滤的推荐算法,该算法可以根据用户特征的相似度找到拥有共同喜好的人,或者根据视频的相似度,找到和用户喜欢的视频相似的其他视频,从而可以很好地减少应用内信息的冗余,获取到用户更为感兴趣的内容,给用户带来更好的使用体验。
2 系统设计
本文将基于个性化推荐系统设计一款视频App。系统的总体架构图如图1所示。……