Analyzing Communication Interaction Networks (CINs) in enterprises and inferring hierarchies

摘要: With the proliferation of electronic modes communication (e.g., e-mails, short messages), employees inside an enterprise can form several distinct Communication Interaction Networks, or CINs for short. A CIN is essentially a graph representation ''who talks to whom'' among group individuals. In this paper, we conduct empirical study two modern enterprises and focus on three main questions: (Q1) How from look; (Q2) use different available within enterprise; (Q3) By only using CINs, how much information extract regarding hierarchy in enterprise. We address these questions Enron Corporation provider, exchange phone-calls, messages (SMS). For Q1, reveal following key structural properties that are shared by all our study: they have high edge density, clustering coefficient, close zero assortativity coefficient. Q2, observe differences various modes. This suggests capture behavioral Q3, propose HumanRank, method ranking individuals based their importance CEOs having higher rank than ordinary employees) interactions between them. Next, introduce unsupervised parameter-free algorithm identifies hierarchies separating managers employees. Our achieves above 70% accuracy outperforms state-of-the-art [23].