Behavioral Targeting Advertising Research Based On LDA

With the development of e-commerce, online media advertising become one ofthe main sources of revenue to web site, constantly changing online advertisingtechnology has led to the ads without purpose, the ads present to user is almost theyare not interested, not only does not cause the user’s interest away to click advertising,but also lead to the user’s bored.In response to this situation, there has been a precisethe concept of advertising.Online advertising development trend of precision led tobehavioral targeting online advertising technologies, such advertising models withprecision, timeliness, efficiency, thus drawn the people’s attention. Behavioraltargeting advertising works by Mining user browsing behaviour of users contained inthe record for advertising, according to the user’s behavior, advertisers can moreaccurately target users for advertising, with a minimum investment for maximumprofit. Due to the wide coverage, high precision targeted ads feature being the majorInternet companies and valued by precision advertising service providers.This article first introduces background and significance of behavioral targetingonline advertising research, characteristics of behavioral targeted advertising and puton the difficulties in the process, leads to the research and innovation of this paper,then introduced the status of precision ads technology.Discusses the pretreatmentprocess of a Web page, document clustering methods, and common statistical modelof the document. Then, for behavioral targeting advertising, presented a study onbehavioral targeted advertising based on LDA algorithm.The algorithm based on user history to access records, introduced the concept oftracking window, arranged according to the order of Web page access time, theestablishment of time-varying model of user behavior using an improved k-meansalgorithm to calculate the user’s long-term and short-term behaviour; and then by thetime-varying model of user behavior with LDA theme model, presents an efficientalgorithm of behavioral targeting advertising LDA_M. Finally based on the LDA_Malgorithm is most representative of the user matches several keywords and advertisingkeywords elect and behavioral characteristics of users with the most relevant ones to put ads. The algorithm solves the traditional advertising method can not analyzesemantics and LDA model advertising mining algorithms are not well drawback userbehavior change. Experiments show that the algorithm can distinguish very well theuser’s long and short term, precisely Mining user’s behavioral characteristics, put inuser behavior-related advertisements, improves the accuracy of the ads.