Nowadays, online advertising is regarded as the most prospective form of advertising. One kind of online advertising is banner advertising. However, click through rate (CTR), which measures banner effectiveness, has been declining as low as 0.1 percent. CTR is important for publishers since the payment from advertisers is based on the number of clicks. Previous studies show that showing internet users with banners that match their online behavior will increase CTR. This behavioral targeting can be implemented by classifying the users based on their click-stream data from their navigation and present the banners to users whose history behavior indicates high interest on advertised products. This paper proposes a web ad selector model in order to determine suitable banner advertising to the audience using their browsing history (web server log file). This web ad selector consists of two models: fuzzy inference model and web ad matching model. From the face validity our web ad selector model always includes the banner ad selected by the expert, with matching score average of 75 percent. This proves that our proposed model has a good performance.