The Research Of Targeted Advertising Algorithm

With the rapid development of the Internet advertising industry, targeted advertising as a new network advertising model is also developing very quickly. This kind of advertising is very attractive because it's accurate, timely and efficient. Targeted advertising is a kind of advertising which is delivered on web pages with the page content or user behavior related. Targeted advertising can be divided into content targeted advertising and behavioral targeting advertising according to the different targets. For the content targeted advertising, we present an efficient semantic-based content targeted advertising algorithm. This algorithm first of all uses the semantic relations between words to expand web page keywords and calculates the weights of these expanded words with the help of Approximate Bayesian Weight Computing Model, then computes the similarities between expanded web pages and ads using the Improved Vector Space Model. At last we sort ads according to these similarities and find proper ads for web pages. For the behavioral targeting advertising, we propose a new behavioral targeting advertising algorithm which is based on the user behavior feature analysis. The algorithm first of all clusters the users' recently visited web pages according to the User Behavior Feature Model, then analyzes each cluster of web pages using the User Behavior Feature Analysis Algorithm and calculates its weight accordingly. With this weight and the similarity between the cluster center and ads, we can calculate the final scores of ads. At last we sort ads according to their scores and find proper ads for the users. A large number of experiments show that these two algorithms are very effective targeted advertising algorithms which can match ads and the web page content or user behavior effectively and have a very bright future.