Feasibility of Extracting Unique Signature for Each User Based on Analysing Internet Usage Behaviours

this paper presents the results of our study about feasibility of extracting unique web signature for each user by exploring Internet usage behaviors. Proxy server access log files of one engineering college are collected for 32 months continually for this study. Behaviors of users are compared with each other in various aspects including their time spend on Internet per day. Further the relationships between visited websites by users and time of connection per day, relationships between number of hits (number of visited webpages in a day) and time of connection and gender of users is explored. The results of this study confirm the feasibility of having unique web signa-ture for users and the length of extracted web signature in our study was 1 to 78. Approximately for 82% of total users in our study, we extract unique web signature. In an average analyzing the access log files of proxy servers for 15-20 continues days is sufficient for extracting signature of users. K-means and DBSCAN clustering methods are used for creating clusters of users based on the similarity of their web signatures. Some applications of Web-signature are: • Help to administrators for predicting the more overload timing per day and plan for that. • Grouping users based on their similarity of contents of Web-Signatures and establishing intelligent social network between those users based on their similar behaviors.

Web Signature;Internet Usage Behavior; Access log files; Proxy Server,;k-means; DBSCAN