Document Type



One of the widely used methods for product recommendation in Internet shopping malls is matching product features against customers’ profiles. In this method, it is very important to choose suitable set of features for recommendation efficiency and performance, which has, however, not been rigorously researched so far. In this paper, we build a data set collected from a virtual Internet shopping experiment and adapt and apply feature reduction techniques from pattern matching and information retrieval fields to the data to analyze recommendation performance. The analysis shows that the application of SVD (Singular Value Decomposition) can be the best among the applied methods for recommendation performance.