Internet Advertising Precise Delivery Platform Based On User Behavior Research

With the rapid development of online advertising industry, ad delivery efficiency advertisers increasingly become the focus of attention at the precise needs of advertising promotion, targeted advertising this new advertising model has also been a rapid development of this kind of advertising its precise mode of delivery, timely and efficient features more and more people of all ages. Targeted advertising is essentially a personalized recommendation technology for online advertising technology to build user interest feature model by mining user browsing behavior and browse the contents and put its interest ads based on user interest model.This paper presents an efficient features based on user browsing behavior of behavioral targeting advertising algorithm that analyzes the user behavior characteristics, and web browsing features, different behavioral characteristics impact on user interest is different, based on characteristics of the on the traditional vector space model is improved so that it can more accurately express the user interest. In order to alleviate the problem of semantic-based user interest model mismatch and text keyword advertising representation between semantic keywords associated tools can be extended and improved algorithm based on keywords Chinese Wikipedia classification system and link structure and extended semantic relevance, then the use of Bayesian network model to calculate the weight to calculate the right words in the heavy expansion. Then introduces the characteristics of user behavior model, user behavior into short-term and long-term behavior, introducing the concept of the sliding window and one of the browse documents by topic clustering, the weight of each effective cluster computing clusters formed heavy, put behavior freshness and dispersion behavior is used to distinguish the concept of long-term and short-term behavior characteristics behavioral characteristics, combined with effective cluster weights and cluster centroid similarity advertising gives advertisers score calculation.