Using Self Organizing Feature Maps to acquire knowledge about visitor behavior in a web site Juan D. Vel´asquez, Hiroshi Yasuda, Terumasa Aoki 1 , Richard Weber 2 , and Eduardo Vera 3 1 Research Center for Advanced Science and Technology, University of Tokyo {jvelasqu,yasuda,aoki}@mpeg.rcast.u-tokyo.ac.jp 2 Department of Industrial Engineering, University of Chile, rweber@dii.uchile.cl 3 AccessNova Program, Department of Computer Science, University of Chile, esvera@accessnova.cl Abstract. When a user visits a web site, important information con- cerning his/her preferences and behavior is stored implicitly in the asso- ciated log files. This information can be revealed by using data mining techniques and can be used in order to improve both, content and struc- ture of the respective web site. From the set of possible that define the visitor’s behavior, two have been selected: the visited pages and the time spent in each one of them. With this information, a new distance was defined and used in a self organizing map which identifies clusters of similar sessions, allowing the analysis of visitors behavior. The proposed methodology has been applied to the log files from a certain web site. The respective results gave very important insights regarding visitors behavior and preferences and prompted the reconfiguration of the web site. 1 Introduction When a visitor enters a web site, the selected pages have direct relation with the desired information he/she is looking for. The ideal structure of a web site should support the visitors in finding such information. However, reality is quite different. In many cases, the structure of a Web site does not help to find the desired information, although a page that contains it, does exist [3]. Studying visitors behavior is important in order to create more attractive contents, to predict her/his preferences and to prepare links with suggestions, among others [9]. These research initiatives aim at facilitating web site navigation, and in the case of commercial sites, at increasing market shares [1], transforming visitors into customers, increasing customers loyalty and predicting their preferences. Each click of a web site visitor is stored in files, known as web logs [7]. The knowledge about visitors behavior contained in these files can be extracted using data mining techniques such as e.g. self-organizing feature maps (SOFM).