A Hybrid Approach for Knowledge-Based Product Recommendation

Abstract

Knowledge-based recommendation has proven to be useful approach for product recommendation where individual products are described in terms of well defined set of features. However existing knowledge-based recommendation systems lack on some issues: it does not allow customers to set importance to the feature with matching options and, it is not possible to set cutoff at individual feature as well as case level at runtime during product recommendation process. In this paper, we have presented a hybrid approach, which integrates rule based reasoning (RBR) and case based reasoning (CBR) techniques to address these issues. We have also described how case based reasoning can be used for clustering i.e. identifying products, which are similar to each other in the product catalog. This is useful when user likes a particular product and wants to see other similar products.