Using back-propagation to learn association rules for service personalization

A novel, practical and efficient strategy for providing personalization for online shoppers is proposed. Based on shoppers’ previous purchasing behavior and the customer’s previous choices the strategy is capable of suggesting relevant and desirable products to each customer accurately. The strategy is based on training a back-propagation neural network with association rules that are mined from a transactional database. Unlike most strategies that only consider the relationship between the purchased items, the proposed strategy also incorporates additional influential attributes such as the price of items and their merchandise category. A powerful confidence estimate is used to rank the suggestions. Experimental evidence is provided to demonstrate the effectiveness of the proposed strategy.