个性化广告推荐系统及其应用研究

Research on Personalized Advertising Recommendation Systems and Their Applications

近年来,随着互联网及智能移动设备的发展和普及,丰富了广告的推送方式和投放平台.但是传统的广告推送无法满足用户对个性化广告的需求,导致用户对广告产生抵触情绪,给广告推送带来极大的挑战.个性化广告推荐系统作为应对这些挑战的有效手段,成为个性化服务领域的研究热点之一.个性化广告推荐系统获取用户兴趣偏好,利用多种个性化广告推荐技术,通过PC端、移动终端等多平台为用户提供个性化广告,并且已经在一些应用系统中取得不错的效果.本文对个性化广告推荐系统的研究进展进行系统地综述,从个性化广告推荐的概述出发,对近年来个性化广告推荐的关键技术进行深入分析,包括数据采集与预处理、用户偏好获取、个性化广告推荐技术等.统计分析了个性化广告推荐中使用的多种数据集和评价指标,总结当前个性化广告推荐在传统互联网、移动服务、数字标牌、IPTV等场景下的应用.最后对个性化广告推荐系统存在问题和未来深入研究的方向进行讨论和展望.

In recent years,with the development and popularization of the Internet and smart mobile devices,the promotion methods and delivery platforms of advertisements have enriched.However,the traditional advertising push can’t meet the needs of users for personalized advertising,resulting in users’resistance to advertising,which brings great challenges to advertising push.As an effective means to deal with these challenges,personalized advertising recommendation system has become one of the hottest research topics in the field of personalized services.Personalized advertising recommendation system obtains users’interests and preferences,uses a variety of personalized advertising recommendation technologies,provides personalized advertising for users through PC-side,mobile terminals and other platforms,and has achieved good results in some application systems.Therefore,personalized advertising recommendation system can alleviate the influence on users caused by the content of advertisements does not meet the needs of users,or the time or location of advertisements is not appropriate.At the same time,personalized advertising recommendation system can effectively deal with the challenges brought by advertisement recommendation due to the negative emotions of users and the decrease of advertising profits,and reduce users’resistance to advertisements and enhance user experience,which not only meets users’personalized needs,but also considers the economic benefits generated by advertisements.This paper summarizes the research progress of personalized advertising recommendation system.Starting from the introduction of personalized advertising recommendation,this paper expounds the differences between this paper and the review of the existing advertising recommendation system,and leads to the writing motivation of this paper.It points out the similarities and differences between personalized advertising and non-personalized advertising such as other forms of advertising and traditional advertising,analyzes the differences between personalized advertising recommendation and other personalized recommendation,highlights the uniqueness of this field.Then,this paper deeply analyses the key technologies of personalized advertising recommendation,including data acquisition and preprocessing,acquisition techniques for explicit and potential user preferences,collaborative filtering,context,hybrid recommendation,click rate prediction and other personalized advertising recommendation technologies,and classifies these key technologies in detail,finds their advantages and disadvantages,and draws the direction of improvement.Subsequently,this paper makes a statistical analysis of the internal data set,the open data set and the data set captured by some researchers in personalized advertising recommendation.After that,it summarizes the traditional evaluation indicators and specific evaluation indicators commonly used in advertising recommendation,and makes a comparative analysis and explanation of some studies combining with the main data sets and evaluation indicators.Then it organizes the current application of personalized advertising recommendation in traditional Internet,mobile service,digital signage,IPTV and other scenarios,and finds out the future development direction of each scenario,and some typical applications are listed and analyzed.In addition,advertising recommendation needs to be improved in the aspects of the timeliness of advertising recommendation,privacy protection of users,cold start in advertising recommendation,content sensitivity of advertising recommendation,dynamic preference acquisition of users,extension of content-based advertising context recommendation,localization of mobile advertising recommendation system,and utilization of multi-source data.Finally,in order to develop the personalized advertising recommendation system,this paper discusses and prospects the problems and the further research in the future in the above aspects of advertising recommendation.