Instant Messaging is a popular medium for both social and workrelated communication. In this paper we report an investigation of the effect of interpersonal relationship on underlying basic communication characteristics (such as messaging rate and duration) using a large corpus of instant messages. Our results show that communication characteristics differ significantly for communications between users who are in a work relationship and between users who are in a social relationship. We used our findings to inform the creation of statistical models that predict the relationship between users without the use of message content – achieving an accuracy of nearly 80% for one such model. We discuss the results of our analyses and potential uses of these models. Categories and Subject Descriptors H5.2 [Information interfaces and presentation]: User
Daniel Avrahami, Scott E. Hudson