On using query neighbourhood for better navigation through a product catalog: SMART approach

We present an approach for efficient navigation through an on-line product catalog. The goals of the approach are (i) to support a user in a step-by-step searching for a product and (ii) to ensure the product he has found is the best solution for his need. The product database is based on a light-weighted ontology and structured in the form of a lattice that enables hierarchicals clustering of users' requests. It means that a user can navigate from a very general product-attribute cluster, which satisfies user's buying need and contains a lot of products, to the very specific clusters that seem to contain a few, but for the user highly relevant, products. In each navigation step, the user is provided with the minimal set of navigation alternatives. In order to enhance the processing of the lattice, the partial ordering of the products-attributes clusters is computed using methods from formal concept analysis.