For Want of a Bit the User Was Lost: Cheap User Modeling

398 The more a computer knows about a user, the better it can serve that user. But there are different styles, and even philosophies, of how to teach our computers about us—about our habits, interests, patterns, and preferences. " Cheap " user modeling, the subject of this essay, simply means ascertaining a few bits of information about each user, processing that information quickly, and providing the results to applications, all without intruding upon the user's consciousness. In short, there are techniques for personalization that can—and should—be built into today's systems. Like most journal papers, this is a description of an existing system: DOPPELGÄNGER. But it is also an exhortation for readers to incorporate the described techniques and philosophy into their own systems. ser modeling is nothing more than a fancy term for automated personalization. Humans model each other all the time. I am modeling you as I write; my topics, presentation, and language are all aimed at a hypothetical, average reader of this journal. If I have guessed well, you will enjoy this essay. If not, you will skip to the next one. That is what user modeling systems do—they make guesses, and hopefully educated ones, about their users. Of course, not all readers are the same, and what entices one reader may alienate another. Ideally, this essay would itself be adaptive, able to gauge the interests of individual readers and transform itself accordingly. But since that is not yet possible, I will employ some " cheap " user modeling. I will make a one-bit distinction between philosophy and engineering in this essay: after this paragraph, the left column contains the " big picture, " and the right column contains details. (The separation is suspended for the section " A user modeling toolbox, " then resumes until the " Reprise. ") Read either column, or both, depending on your interests. ©Copyright 1996 by International Business Machines Corporation. Copying in printed form for private use is permitted without payment of royalty provided that (1) each reproduction is done without alteration and (2) the Journal reference and IBM copyright notice are included on the first page. The title and abstract, but no other portions, of this paper may be copied or distributed royalty free without further permission by computer-based and other information-service systems. Permission to republish any other portion of this paper must be obtained from the Editor.