The notion of proximity for innovation clusters needs to expand compared to traditional agglomeration literature because it involves multiple dimensions such as geography, business, and knowledge; however, limited research probed into thisit requires multidimensional data and novel methods. This article, therefore, proposes a three-layered framework that uses multisource heterogeneous data and network methods to measure the cluster-proximity in innovation clusters, in order to understand better the combined proximity between organizations within/across network communities. First, we developed a threelayered framework to map the network-based innovation cluster by integrating patent citation, business transaction, and geographic data. Second, in the innovation cluster, we identified the network communities in knowledge, business, and geographic layers, respectively. Third, we measured the cluster proximity within/across communities by using a combined index. We selected A City's machine tool sector in China as a case. This article found that machine-tool firms/organizations have higher cluster proximity within geographic communities that are enriched mostly by business connections. In comparison, they have lower cluster proximity across geographic communities, but the proximity is enhanced both by business connections and knowledge linkages. This may imply that knowledge linkages are more important in across-community proximity, and this needs more policy attention.