In this paper, we propose an approach for supporting e-commerce buyers who do not have clear goals and well-defined preferences regarding the products of purchase. This is particularly the case when the purchase of such products is infrequent, e.g. home theaters and laptop computers. Our method allows customers to express vague preferences and dynamically gives advisory information without limiting customer's opportunity to consider all possible solutions. The system is based on fuzzy logic algorithm, cluster analysis, and the well-known 'divergence/convergence' principle from problem solving research. The results of an experiment with the prototype systems are in favor of our proposed methods compared to traditional catalog method.