Masking of actual communication pattern by a queuing mechanism

Abstract

We analyzed the effects of queuing in a system of human activities through the Enron dataset. We found out that the Enron dataset is temporally separated in terms of the servers' method of message management. Here, we show that the sent messages in the Enron dataset was initially processed through queuing, depicted by bursts in Δt which are multiples of 60 seconds, and was converted into a realtime message management at some point. The time of change in the message management was determined in this paper. We argued that the presence of two different sending management in the Enron dataset poses an important implication on researches regarding systems with unknown underlying processes.