Using E-mail Communication Network for Importance Measurement in Collaboration Environments
[ 1 ] Instytut Informatyki, Wydział Informatyki, Politechnika Poznańska | [ P ] pracownik
2014
referat
angielski
EN Can we establish the importance of people by simply analyzing the set of sent and received emails, having no access to subject lines or contents of messages? The answer, apparently, is ”yes we can”. Intrinsic behavior of people reveals simple patterns in choosing which emails to answer next. Our theory is based on two assumptions. We assume that people do their email communication in bursts, answering several messages consecutively and that they can freely choose the order of answers. Secondly, we believe that people use priority queues to manage their internal task lists, including the list of emails to be answered. Looking at timing and ordering of responses we derive individual rankings of importance of actors, because we posit that people have a tendency to reply to important actors first. These individual subjective rankings are significant because they reflect the relative importance of other actors as perceived by each actor. The individual rankings are aggregated into a global ranking of importance of all actors. We perform an experimental evaluation of our model by analyzing the dataset consisting of over 600 000 emails sent during one year period to 200 employees of our university. Our final ranking closely reflects the ”true” importance of employees computed based on surveys. We think that our model is general and can be applied whenever behavioral data is available which includes any choice made by actors from a set of available alternatives with the alternatives having varying degrees of importance to individual actors.
43 - 54
WoS (15)