A Survey of Preference Elicitation

As more information becomes available to users, there is an increasing number of complex tasks that motivate the need for computer-assisted interaction, such as the exploration of a large multidimensional information space. In order to assist users with achieving their goals, computers need to accurately model the preferences of their users. Preference elicitation techniques attempt to collect important information about the user’s preferences and interests. This survey describes various preference elicitation techniques and systems. Furthermore, the survey explores the potential for integrating preference elicitation techniques into an existing assisted navigation system designed to aid users with data exploration. The navigation system assists users by identifying and locating interesting elements within the visualization of a multidimensional dataset. The survey will also discuss how the incorporation of preference elicitation techniques could potentially benefit the navigation assistant. Figure 1: A visualization of a weather dataset composed of perceptual texture elements (or pexels). Dense areas represent high cloud cover, irregular areas represent high precipitation, and red and pink colors correspond to high temperature values. Elements of interests have been identified and clustered (shown in blue). The underlying navigational graph framework is also shown.