A Study on Traceability Data-Mining

可追溯性数据挖掘是一种分析在装配过程中哪些零件导致产品级故障的技术，例如汽车制造，其中超过万个零件由多个供应商以分层方式装配。在这项研究中，我们引入了虚拟地段的概念，并开发了钻取挖掘方法，可以有选择地搜索可疑零件。在实现该系统后，我们将我们的方法与现有的方法进行了比较，结果表明，即使数据量增加，我们的方法也具有可行的搜索时间。

Traceability data-mining is a technology to analyze which of the parts contributes to the product-level fault in the assembly processes such as automobile manufacturing, where more than ten-thousands of parts are assembled by multiple vendors in hierarchical manner. In this study, we introduced a virtual lot concept and developed the drill-down mining method which can search the suspicious parts selectively. After implementing the system, we compared our method with the existing one and showed that our method has a feasible search time even if data size is increased.