基于改进TF⁃IDF的电梯传媒广告推荐方法  

Recommendation method of elevator media advertising based on improved TF⁃IDF algorithm

作　　者：陈彦彬[1,2] 杨泽华 薛晓桂 黄锦钿 Chen Yanbin;Yang Zehua;Xue Xiaogui;Huang Jintian(Training and Information Center,Jieyang Polytechnic,Jieyang 522051,China;Engineering technology Research Center,Guangdong Bohua Technology Co.,Ltd.,Jieyang 522000,China;Department of information Engineering,Jieyang Polytechnic,Jieyang 522051,China;School of Physics and Electrical Engineering,Hanshan Normal University,Chaozhou 521000,China)

机构地区：[1]揭阳职业技术学院实训与信息中心,揭阳522051 [2]广东博华科技有限公司工程技术研究中心,揭阳522000 [3]揭阳职业技术学院信息工程系,揭阳522051 [4]韩山师范学院物理与电子工程学院,潮州521000

出　　处：《现代计算机》2023年第8期34-39,共6页Modern Computer

基　　金：广东省科技厅驻镇帮镇扶村农村科技特派员重点派驻任务(KTP20210400);广东省普通高校特色创新项目(2018KTSCX139);揭阳市科技计划项目(2019070)。

摘　　要：随着国家个人信息保护法的出台,电梯传媒终端精准广告投放面临多方面挑战,如何在不采集市民隐私信息的情况下,提高电梯传媒终端广告投放的精准度,为广告投放商提高经济效益是当前终端计算广告研究的重点,为此提出了基于改进TF⁃IDF的电梯传媒广告推荐方法。利用改进的TF⁃IDF对电梯点周边POI、居民等情况进行标签提取,构建了电梯传媒终端标签向量模型;然后利用商户对电梯的评分计算商户对标签兴趣度,最后构建商户兴趣模型对TOP⁃N部电梯终端广告位进行排序,推荐给商户。实验结果表明,该推荐方法的准确率、召回率等均优于采用传统TF⁃IDF算法的结果,而且不用采集市民个人隐私数据,具有较强推广应用效益。With the introduction of the National Personal Information Protection Law,accurate advertising of elevator media terminal faces many challenges.How to improve the accuracy of elevator media terminal advertising without infringing on the privacy of citizens,and improve the economic benefits for advertising providers is the current focus of terminal computing advertising research.Therefore,a recommendation method of elevator media advertisement based on improved TFIDF is proposed.The improved TFIDF was used to extract the labels of POI and residents around the elevator points,and the label vector model of the elevator media terminal was constructed.Then use the merchant’s score to calculate the user’s interest in the label.Finally,a merchant interest model is constructed to sort the data of TOPN elevators and recommend them to merchants.Experimental results show that this recommendation method is superior to the traditional TFIDF algorithm in accuracy and recall rate,and does not need to collect citizens’private data,which has strong promotion and application benefits.

关 键 词：词频-逆文本频率指数 传媒广告 标签评分 用户兴趣 个性化推荐 

分 类 号：G206[文化科学—传播学] F713.8[经济管理—广告] TP391.3[经济管理—产业经济]