基于商业WiFi广告个性化推荐算法的研究

发布时间：2018-01-10 08:30

本文关键词：基于商业WiFi广告个性化推荐算法的研究　出处：《北京交通大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着移动互联网的迅猛发展和智能终端的普及,人们对于无线网络的需求越来强烈,WiFi是移动互联网的底层流量入口,作为公众WiFi的一部分商业WiFi的出现成功弥补了其覆盖范围不足的缺陷,在一定程度上满足了用户联网的需求,同时作为用户上网的底层入口,具有很大的潜在价值。现如今商业WiFi无线认证界面只是简单的横幅广告或者文字广告,页面内容变化还需靠人工来完成,所以更新率低,最重要的是无法根据用户的喜好进行个性化推荐,做不到“千人千面”。一方面用户的体验度不好,另一方面商场中有那么多的商品无法做到精准营销,不利于将流量变现。因此研究在商业WiFi环境下广告个性化推荐算法具有重要意义。本文的主要研究工作如下:(1)对目前主流的推荐算法进行了深入研究,分析了各个算法的推荐原理,并通过配图对原理进行了补充。重点分析了协同过滤算法,认真学习了 Cosine、Pearson以及Adjust Cosine相似度算法,详细描述了基于项目的协同过滤算法的执行过程;(2)提出并分析了一个新的基于关联的相似度算法。通过研究分析商业WiFi环境下用户和广告特征,在协同过滤算法的基础上,提出了一种新的相似度算法,该算法引入了用户/项目的关联因子。研究并优化了评分预测算法,在基于项目的协同过滤算法基础上,对评分预测算法进行了改进,推荐准确度更高,使其更适合商业WiFi的广告环境;(3)通过MAE(平均绝对误差)与RMSE(均方根误差)值来计算文中算法的预测精确度,完成了基于MovieLens数据集的实验,结果证明了新的相似度算法和改进后的评分预测算法在数据不同稀疏程度、不同最近邻个数等情况下,对比传统协同过滤算法具有更好的推荐准确度;(4)最后,利用本文的算法思路,设计了商业WiFi认证广告系统框架,分析了客户端与服务器进行通信的网关协议,利用当前的开源路由器固件OpenWRT和认证服务器固件OpenWMS,设计并实现了商业WiFi的接入认证模块。
[Abstract]:With the popularity of the rapid development of mobile Internet and intelligent terminals, demand for wireless network more and more, WiFi is the underlying traffic entrance of the mobile Internet, as the public part of the WiFi business WiFi make up the defect of inadequate coverage, to a certain extent to meet the needs of Internet users at the same time, the bottom entrance as Internet users, has great potential value. Now the business of WiFi wireless authentication interface simply banner or text ads, the content of the page changes also need to rely on manual to complete, so the update rate is low, the most important thing is not according to the user's preferences for personalized recommendation, do not "Thousand Faces" on the one hand. The user experience is not good, on the other hand, there are so many shopping malls in the goods can do precision marketing, is not conducive to the flow is realized. So the study on commercial WiFi Under the environment of advertising personalized recommendation algorithm has important significance. The main research work of this paper is as follows: (1) the current mainstream recommendation algorithm has been studied, analyzed the principle of each recommendation algorithm, and by drawing on the principle of complementary. Focus on the analysis of the collaborative filtering algorithm, to seriously study the Cosine, Pearson and Adjust Cosine similarity algorithm is described in detail the implementation process of the collaborative filtering algorithm based on project; (2) proposed and analyzed a new similarity algorithm based on correlation analysis. Through the study of households and commercial advertising features under the environment of WiFi, based on collaborative filtering algorithm, proposes a new similarity algorithm. The algorithm introduces the correlation factor user / project. Research and optimization of the score prediction algorithm based on collaborative filtering algorithm based on the project, the score prediction algorithm for the improved push Recommend more accurate, to make it more suitable for commercial WiFi advertising environment; (3) by MAE (mean absolute error) and RMSE (mean square error) value to calculate the prediction accuracy of the algorithm, completed the experiments based on MovieLens data set, the results proved that the new similarity score calculation method and prediction algorithm the improved sparse data in different degree, different nearest neighbor number of cases, compared with the traditional collaborative filtering algorithm has better recommendation accuracy; (4) finally, using the idea of this algorithm, the design of commercial advertising system WiFi authentication framework, gateway protocol analysis of client and server communication, using the current open source router firmware OpenWRT and authentication server firmware OpenWMS, the design and implementation of access authentication module of commercial WiFi.

【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.3;TN92

【相似文献】