Evolving User Pro les to Reduce InternetInformation

This paper discusses the use of Evolving Personal Agent Environments as a potential solution to the problem of information overload as experienced in habitual Web surrng. Some rst experimental results on evolving user prooles using speciating hybrid GAs, the reasoning behind them and support for their potential application in mobile, wireless and location aware information devices are also presented. 1 Information Overload In everyday life, the Internet user is faced with the ever increasing problem of information overload, whether this occurs at home, at the workplace, or as will soon be happening, everywhere 1] 2]. The overwhelming information feed that computer users face leads to anxiety, strain, ineeciency and nally results in uninformed (or misinformed) and frustrated users 3] 4]. Continuously and increasingly Internet users are confronted with laborious and diicult tasks of information ltering and/or gathering, which are inherently computer-oriented processes and need not involve human labour. Moreover, the eeciency and speed with which computer systems may retrieve and process massive amounts of data actually exacerbates the problem of ineeectiveness in information ltering without user intervention. In our research programme, which we rst presented in 5], we aim to solve the information overload problem arising from habitual surrng, with a focus on mobility, Wireless Information Devices (WIDs) and privacy. Therefore moving away from the model of what is of interest to the user to the model of what is of interest to the user here and now. In 5], we surveyed recent eeorts at tackling the problem of information overload. Our view is that, as a means of achieving our goal, we need to research Evolving Personal Agent Environments (EPAE). The focus in this paper is on evolving individual user prooles, which form a vital part of the EPAE research. Thus we aim to describe and explain our deenition of user prooles in personal agent environments, and to support the 2 John Pagonis and Mark C. Sinclair motivation behind it. We also include an overview of our proposed research programme on EPAE and present some early experimental work on user proole evaluation and evolution.