Extracting user posting behavior using HTTP flow

We have proposed a decision-tree-based method for quick extraction of user posting behavior from flow data. Since our method scans only HTTP POST headers fields, which are described at the beginning of the HTTP body, we do not have to fully scan the HTTP body. Thus, traffic flow can be quickly scanned to extract posting behavior. However, it might be difficult to improve the F-measure based on HTTP header fields. Finally, required accuracy is determined by services that use posting behavior; therefore, we need another method, which can be combined with our proposed method for analyzing user behavior using the HTTP body. So far, we assume behavioral targeting advertising (BTA) and Q&A sites as services using posting behavior. Our proposed method might be used for these services. For BTA services, it might help in finding people who are better consumers. For Q&A services, it might help in finding people who are better answerers.