Visualizing Recommender System Results via Multidimensional Scaling

Web site visitors who look for desired items can formulate search queries which are taken by recommender systems to provide support within the underlying buying situation (e.g., enabling users to view recommended items and buy the ones they find most appropriate). Data from a large German retail online store is used to visualize products viewed most frequently together with search profiles that represent identical search queries of larger subgroups of site users. Comparisons between products viewed most frequently and those purchased most frequently can be used to improve the generation of recommendations. The results give interesting insights concerning the searching, viewing, and buying behavior of online shoppers.