Exploration space of human-data interaction

Official URL: http://risc01.sabanciuniv.edu/record=b1669823 (Table of Contents)

Data is everywhere. Starting with the invention of writing, representation artifacts brought the data to observable state which led to natural establishment of an interaction form between human and data. In the human-data interaction (HDI) environment, data representations and analytic systems act as an intermediary role. I suggest a new de nition for HDI in which this interaction is conceptualized as a communication model over a set of media. The interaction occurs with the exchange of messages originated from both human and data. Timing and content of the messages are employed to facilitate objective evaluation of properties of analytic system in question. To systematically investigate the complex nature of HDI, my methodology postulates the phenomenon as a high-dimensional space in which data analytic systems could be positioned based on their properties. Evaluation of the properties are performed based on solid de nitions of the dimensions. I de ne ve properties for data analytic systems, namely, responsiveness, communication media level, unit task diversity, closeness factor, and progressiveness level, and demonstrate how these properties could be objectively calculated. I visually explore the HDI space in which data analytic systems reported in my thesis are plotted on a two-dimensional Cartesian system whose axes are responsiveness and communication media level. Visually identi able patterns in this plot, which I call realms, are characterized by quantitative and qualitative analysis of objective, behavioral, and subjective data collected during the user interaction with the corresponding analytic system.