Data-driven Context Modeling: Review of the State of the Art

In requirements elicitation of context-aware functionalities, context modeling is an early step responsible for understanding what is the context for a certain application domain and how the context influences user tasks of interest. In practice, it has been overlooked, though: Context modeling activities are perceived as time-consuming, non-intuitive, and error-prone. In a scenario with dozens of contextual elements (such as time, location, user characteristics, among several others), the number of possible combinations of contextual elements that may influence user tasks is too high to be analyzed manually. To cope with this problem, data-driven approaches seem to be promising: By means of automating context modeling through data mining, the identification of which contexts influence user tasks of interest can be facilitated. In this document, we report on our literature review of existing work in the field of data-driven context modeling to support requirements elicitation of context-aware functionalities.