Women take relatively longer time for shopping comparing with men. This paper presents a system for product retrieval and recommendation based on psychology of women's shopping activity, and an implementation of the system for apparel products. This system supposes products which pre-defined keywords are assigned, and icons representing the combination of the keywords. It acts real shopping behavior by displaying a set of the icons associated to the products. The system intuitively displays the distribution of products by closely placing the icons sharing common keywords. Also, it features a product recommendation technique based on user interfaces for elimination and narrow down operations, and an evolutionary computation which adjusts the amount of recommendation. The system well assists the shopping behavior of women who are interested in various products.