抄録
Among various corporate marketing activities, the approach to customers is becoming more
sophisticated year by year. In particular, the recommendation with purchases related data has become a
common technology on EC sites as the Internet spreads. However, effective recommendations are not easy in
EC because of the problems of basket dropping and cancellation. In this paper, we discuss a recommendation
method that considers post-purchase cancellation using purchase-related data. A Bayesian network that can
consider probabilistic causal relationships is used for this recommendation analysis.
