Lately, developing systems fully interoperable with others is considered an essential element for successful projects, as not only do e-commerce becomes ubiquitous but also distributed systems' paradigm spreads. However, since definitions of interoperability vary by viewpoints, it is still difficult to have the same understanding and evaluation criteria on interoperability. For instance, various interoperability parties in military use different definitions of interoperability, and its T&E is not conducted according to the definition, but only to levels of information exchange. In this paper, we proposed a new definition of interoperability as followsm First of all, we collected existing and various interoperability definitions, extracting key components in each of them. Second, we statistically analyzed those components and applied the association rules discovery in data mining. We compared existing interoperability definitions to ours. From this research, we found associations among the components from various definitions applying market-basketanalysis, redefining interoperability. Key findings of this research can contribute to a unified viewpoint on the definition, level, and evaluation items of interoperability.