Heterogeneous Graphlets

In this article, we introduce a generalization of graphlets to heterogeneous networks called typed . Informally, are small induced subgraphs. Typed generalize rich as they explicitly capture the higher-order connectivity patterns in such networks. To address problem, describe general framework for counting occurrences graphlets. The proposed algorithms leverage number combinatorial relationships different For each edge, count few graphlets, and with these counts along relationships, obtain exact other o (1) constant time. Notably, worst-case time complexity approach matches best known untyped algorithm. addition, lends itself an efficient lock-free asynchronous parallel implementation. While there no existing methods has been some work that focused on computing much simpler notion colored graphlet. experiments confirm our is orders magnitude faster more space-efficient than Unlike take hours networks, takes only seconds large millions edges. since graphlet (and graphlets), various can be combined give rise new opportunities applications