On the link(s) between 'D' and 'A' in Mobile Data Analytics

This paper addresses the issue of (meta) management of the efficiency in the realm of applications relying on Mobile Data Analytics (MODA). We postulate that considering the efficient execution of the tasks of gathering, storing/retrieving and querying spatio-temporal data separately from the efficiency of executing control and decision-making (or, for that matter, even model-selection) algorithms may not be the most beneficial avenue in MODA settings. We argue that an important component is the management of the dynamics of the different tradeoff(s) and, more importantly, linking them when orchestrating the processes of the data management and analytics. Towards that end, we present the DNA2 (Data's Natural Associations with Analytics) hypothesis for coupling the data and control aspects in MODA contexts. We discuss in detail the role of uncertainty, along with some aspects of declarative specifications for merging the reactive and the pro-active behavior in MODA.