Knowledge discovery of customer purchasing intentions by plausible-frequent itemsets from uncertain data

摘要

Many previous studies have focused on the extraction of association rules from transaction data. Unfortunately, customer purchasing intentions tend to be uncertain during the decision making process. That is, they cannot be obtained from business transaction data. Therefore, the research problem is how to discover frequent itemsets from uncertain data. This study first proposes a new model to represent consumer uncertainty during the decision making process. This representation scheme is based on possibility distributions. The possibility theory provides an excellent framework for handling uncertain data. In addition, an algorithm is developed to mine plausible-frequent itemsets from uncertain data, which are represented by possibility distributions, and then discover plausible association rules based on these plausible-frequent itemsets. Experimental results show that the proposed model can discover interesting plausible-frequent patterns from survey data which represent customer purchasing decisions.