Induction of compact decision trees for personalized recommendation

We propose a method for induction of compact optimal recommendation policies based on discovery of frequent itemsets in a purchase database, followed by the application of standard decision tree learning algorithms for the purposes of simplification and compaction of the recommendation policies. Experimental results suggest that the structure of such policies can be exploited to partition the space of customer purchasing histories much more efficiently than frequent itemset discovery algorithms alone would allow. Categories and Subject Descriptors H.2.8 [Database Management]: Database Applications— Data Mining Keywords response modeling, product recommendation, frequent itemset mining

• Type: Conference
• Year: 2006
• Where: SAC
• Authors: Daniel Nikovski, Veselin Kulev
• Keywords: Applied Computing, Frequent Itemset, Optimal Recommendation Policies, Recommendation Policies, SAC 2006