Learning human behavioral profiles in a cyber environment

Fast and efficient delivery of targeted and personalized advertisements has become a staple in E-commerce. The field of market research targeting in online environments has seen extensive research on the topic of customer targeting, but most of the information detailing the process of profiling is closely guarded by the industry. Much of what is available tells that information such as cookies and browsing history are used to target customers, associating keywords with specific groups of inventory. We designed and conducted a web-based shopping experiment with fifty participants to observe how people of different backgrounds and experiences make decisions. The experiment investigates whether or not it is possible to learn repeatable patterns of behavior through data mining techniques to determine the profile of an online shopper solely based upon outside observation. Our results suggest that by recording the actions taken and time spent on each decision, one may be able to use statistical models to discern demographic information such as gender and infer connections between the shopper and the gift recipient. This brings optimism to the idea that by examining action data and the target of the actions, our methods could be used to determine the identities of online shoppers. Further, these methods could be tested in other online environments to determine identities, in cases such as online fraud or other issues related to cyber security.