Talk 1: SensePath – Understanding the Sensemaking Process through Analytic Provenance
Sensemaking is described as the process of comprehension, finding meaning and gaining insight from information, producing new knowledge and informing further action. Understanding the sensemaking process allows building effective visual analytics tools to make sense of large and complex datasets. Currently, it is often a manual and time-consuming undertaking to comprehend this: researchers collect observation data, transcribe screen capture videos and think-aloud recordings, identify recurring patterns, and eventually abstract the sensemaking process into a general model. In this talk, I will propose a general approach to facilitate such a qualitative analysis process, introduce a prototype, SensePath, to demonstrate the application of this approach with a focus on browser-based online sensemaking, and present our findings from a preliminary evaluation of the tool.
Phong Nguyen is currently doing his PhD in the Interaction Design Centre at Middlesex University. His research interests include visual analytics, information visualization, with a focus on understanding and supporting the sensemaking process using analytic provenance.
Talk 2: An Empirical study on Perception of Correlation using Scatter Plots
Scatter plots have been in use for over many centuries now but lacks the knowledge of metrics underlying their perception by humans. In this study we empirically assess user performance of estimating correlation in scatter plots for different factors and report whether there is a significant difference and/or relation in the subjective and objective correlation values in relation to different correlation indices, data distribution, symmetry of data enclosure, and number of data points used to plot it. The results suggest that error rates vary in relation to all these factors and condemn the existence of a single linear or non-linear regression pattern to which the human perception of statistical correlation would conform. Even for a set of consistent geometric enclosure such as ellipse itself, the error rate do not conform to a straight line,and thus the possibility of leveraging any perceptual models, such as Weber’s law, to evaluate correlation “accuracy” and “precision” is invalid. These finding are significant in that they are new and publishable and falsifies the conclusion of a recent journal paper where the authors might be misled by the accidental patterns resulting from insufficient sampling of a key experimental variable. Lastly, we also establish that as standalone quantities, both human perception as well as the statistical indicator of correlation are unreliable and need to be considered as a ‘married couple’ which complement each other.
Varshita is a recent graduate from University of Oxford where she was studying Master’s in Computer Science. She earned her undergraduate degree (Bachelors of Science (Honours)) in the same field from University of Delhi, India before moving to U.K. to pursue her masters. Her Master’s research combined psychology with Visual Analytics to question the stability of mathematical models (such as Weber’s law) that quantify human perception.
Her interests include focusing on pan-sector research areas to render visualizations that are more self-explanatory and efficient in context of Big Data along with paying close attention to the social, ethical and legal issues associated with it. She is currently collaborating with Middlesex University where she is serving as a Research Assistant on the project VALCRI (Visual Analytics for sense-making in Criminal Intelligence analysis), post which she will start working on the publication of her paper based on the Master’s thesis which earned her a distinction.