Constituencies for Users: How to Develop them by Interpreting Logs of Web Site Access
The number of electronic journals is growing as rapidly as the World Wide Web on which many are published. Readership of an electronic journal is important to quantify, just as it is for a printed journal. In maintaining the Journal of Interactive Media in Education (JIME), a scholarly electronic journal open to all, we require readership statistics more meaningful than the variations on the theme of "number of hits" given by many log analysis packages. Important aspects of JIME's open access are the decisions to not require subscription, nor the use of hidden tracking aids such as cookies. Readership information is acquired from the server log of client requests. We are investigating the identification of user sessions from such logs, and the development of classification of sessions into constituencies of readership. In this paper, we present the result of manual analysis from which we are developing automatic analysis mechanisms.
AAAI Spring Symposium on Intelligent Agents in Cyberspace, March 22-24, 1999, Stanford University, California