A fast pruning redundant rule method using Galois connection

Besides preprocessing, post-analysis also plays an important role in knowledge discovery. It can effectively assist users to grasp the obtained knowledge. However, many of data mining algorithms merely take performance into consideration and put the post-analysis of results aside. They generate a modest number of rules for the purpose of improving accuracy. Unfortunately, most induced rules are redundant or insignificant. Their presence not only confuses end-users in post-analysis, but also degrades efficiency in future decision task. Thus, it is necessary to eliminate redundant or irrelevant rules as more as possible. In this paper, we present an efficient post-processing method to prune redundant rules by virtue of the property of Galois connection, which inherently constrains rules with respect to objects. Its advantage is that information will not be lost greatly during pruning step. The experimental evaluation shows that the proposed method is competent for discarding a large number of superfluous rules effectively and a high compression factor will be achieved. What’s more, the computational cost of our method is surprisingly lower than the Apriori method.