On Mobile User Behaviour Patterns

We present empirical analysis of human searches for information from mobile devices, focusing on temporal dynamics, semantics, and topics of queries. Our analysis is based on a large scale data of mobile search logs over a week period from a major US mobile service provider. We find that human searches appear in bursts over time with the distribution of the query inter arrival time following a power law decay up to a day and decays exponentially beyond. Interestingly, this finding conforms to some other measures of human activity reported in previous studies. We also provide preliminary characterisation results of the semantics and topics of queries, some of which conform to that of previous studies. The results would be of general interest for understanding the dynamics of human activity and, in particular, may be leveraged for the design of mobile services.