Home buying is a complex process. The needs of home-buyers are idiosyncratic in natures and vary with their demographics and life styles. It takes great effort on part of the home-buyers and real estate agents to ascertain the housing needs. Several online real-estate search systems have become available to help home-buyers and real-estate agents to select suitable houses before they make visits to the properties. But these systems take very simplistic view of home buyers� needs and consider very basic requirements of the buyers. Since a building carries large number of attributes therefore average buyers are unlikely to be able to completely specify their preferences on such a large number of attributes. On the other hand the available systems lack the intelligence to assess the preferences of the buyers without unduly overloading them. As a result they are unable to provide more personalized recommendation to the home-buyers. The question is how we can improve the online real-estate search systems for providing personalized recommendation to the home-buyers. This research proposes to explore a computational framework for identifying the housing preferences of the home buyers for personalized recommendation.