The dynamic HomeFinder: evaluating dynamic queries in a real-estate information exploration system

Reviewer: Donald Harris Kraft

The authors demonstrate a new level of visualization for information retrieval, adding considerably to our understanding of how retrieval systems can and should be used. The focus is on the ability of the user to easily interface with the system, getting immediate visual feedback on the results of each portion of the query as it is entered. This interface is based on direct manipulation techniques and what the authors call a dynamic query. The system makes use of sliders and buttons to indicate choices and ranges of choices to allow users to input their queries. This approach makes it easy to update query choices, while letting the system give immediate feedback on the results of the choices. Of course, the authors note that the screen layout, including color, is set up to provide the maximum in ease of use. The example given in the paper is finding a home in the Washington, DC, area. The initial screen provides a map of the area along with lights that correspond to the locations on the map of the entire database of homes in the prototype system. As users specify on the map where they work and where their spouses work, they can watch lights go on or off as they then specify the maximum distance they will commute to work and the maximum distance their spouses will commute; the minimum and maximum number of bedrooms; the acceptable price range; whether they want a house, apartment, or condominium; and whether they demand a fireplace, a garage, central air conditioning, or a new house (the last choices are buttons, while the others are sliders). Then, looking at any of the remaining lights, the use of the mouse or a touchscreen can cause detailed text about that home to be displayed. The authors discuss some experiments comparing this method with standard database query mechanisms for the advertising text data. The authors' graphical, dynamic query method proves superior in terms of user task speed and user opinion. Novices could learn quickly, query choices are reversible, and no error messages are needed. Of course, the system is not perfect; for example, one is limited in the number and type of fields. In addition, the choices are seen as equally important when evaluating which homes satisfy the user's criteria, but it is obvious that cost may be more important to many people when comparing homes. All in all, the paper is a good read. Ample pictures show the screens so that readers can follow what is being said. Moreover, the paper provides proper detail about an exciting new system and uses an example to show how powerful this approach is.