Mining sequential rules from a sequence database usually returns a set of rules with great cardinality. However, in real world applications, the end-users are often interested in a subset of sequential rules. Particularly, they may consider only rules that contain a specific set of items. The naïve strategy is to apply such itemset constraints into the post-processing step. However, such approaches require much effort and time. This paper proposes the effective methods for integrating itemset constraints into the actual mining process. We proposed two algorithms, namely MSRIC-R and MSRIC-P, to solve this problem in which MSRIC-R pushed the constraints into the rule generating phase, and MSRIC-P pushes into the pattern mining phase. Experiments show that the proposed algorithms outperform the post-processing approach.

从顺序数据库中挖掘顺序规则通常会返回一组具有很大基数的规则。但是，在实际应用中，最终用户通常对顺序规则的子集感兴趣。特别是，他们可能只考虑包含一组特定项目的规则。幼稚的策略是将此类项目集约束应用于后处理步骤。但是，这样的方法需要大量的努力和时间。本文提出了将项目集约束整合到实际挖掘过程中的有效方法。为了解决这个问题，我们提出了两种算法，即MSRIC-R和MSRIC-P，其中MSRIC-R将约束推入规则生成阶段，而MSRIC-P推入模式挖掘阶段。实验表明，所提出的算法优于后处理方法。