User-Oriented Product Search Based on Consumer Values and Lifestyles

Product search engines are essentially unchanged since the inception of online shopping, providing basic browsing by category and “advanced” keyword search. This paper presents a user-oriented product search method based on consumer values and lifestyles that correspond to user purchasing criteria rather than technical specifications. The technique is suited to relatively infrequent purchases where users possess little domain or market knowledge and existing interfaces are difficult to use. We show how to construct a knowledge base to support a user-oriented product search engine without the need for a domain expert to manually label the items. We present Lifestyle Car Finder, a user-oriented product search system in the domain of new cars. The system incorporates various modes of navigation (search refinement, a new form of critiquing adaptive to the user’s query, and breadcrumb trails) and decision support (similar car comparison, explanations and technical specifications). We report on a user study showing that, broadly speaking, users were highly satisfied with the system and felt they were confident in their decisions. 1 User-Oriented Product Search Most online shopping search interfaces provide only basic browsing and “advanced” keyword search using technical features. These interfaces are difficult to use for users without domain knowledge, especially when products are highly technical and/or where the market contains many items that are very similar and thus hard to differentiate. The typical difficulty faced by everyday users is to express their information needs in the manner expected by the search engine. For some systems, users have to know the brand or the technical name of the product (e.g. multi-function printer), however naive users often start with little or no domain knowledge and so need to construct their product and brand preferences through their interaction with the system, Bettman, Luce and Payne [2]. With text-based keyword search interfaces, the user has to enter sufficiently precise keywords for the system to return useful results, however naive users only have a rough idea of their high-level requirements, and do not know how to express their needs in technical terms. Adding to the problem, naive users are often casual users, so there is limited interaction history to tailor the interaction. This motivates the need for systems that enable users to express their requirements using needs-based attributes that are meaningful to the user and in accord with the