Skip to main content

Pair Analytics

 

Pair analytics is a research method in which visual analytic reasoning, collaboration in analytical work and analytic discourse can be captured and analyzed This method is loosely based on Pair Programming from Extreme Programming software development methods. Pair analytics requires a dyad of participants: one Subject Matter Expert (SME) and one Visual Analytics Expert (VAE). These two participants are given one analytical task, one data set, and one computer with a VA tool. Each participant is assigned to a specific role. The VAE plays the role of the “driver” and the SME plays the role of the “navigator” of the VA tool. The VAE should have mastered the operation of a suite of VA tools and have an understanding of cognitive processes that mediate analytic tasks, but may lack the contextual knowledge that would be required to conduct meaningful analysis of the data set s/he is working on. The pairing of VAE and SME in the context of an analytic task is designed to generate a human-human dialog that will make explicit the mental model and analytic cognitive processes of the SME. For example, the SME may use expertise-related schemas to structure her analysis, detect anomalies in the visual representations of the data, suggest alternative analytical paths, and identify the domain-specific value of data patterns/trends found in the analytical session.

There are a number of research products that are expected to result from this procedure. First, we expect that the conclusions reached in the analysis will be superior to those that might be generated though traditional analytic methods. Second both SME and VAE should learn something about each other’s domain of expertise through the process of negotiating a common ground of knowledge needed to drive the joint activity of pair analytics. Analysis of this process can be used to derive instructional content for analysts and toolmakers. Third, the process of negotiating common ground is itself of interest since it makes explicit reasoning processes of interacting with the visual representations of the data. IThe process can be captured and coding strategies developed for its description by video analysis. The research goal here is to develop a theory of psycholinguistic interaction for visual analytics in order to predict effective cognitive performance and to recommend design guidelines and evaluation metrics for visual analytics applications.

This method is being used in partnership with The Boeing Company and Boeing Canada's Aeroinfo to look at analysis of aircraft safety, reliability and maintainability. Funding is provided by the Boeing Company and Aeroinfo (Boeing Canada)  in the form of student internships for work at Boeing and Aeroinfo, and by the SFU/UBC grant "Boeing support for visual analytics in Canada".

Papers:

Arias-Hernandez, R, Kaastra, L.T., Green, T.M. and Fisher, B. 2011, "Capturing Reasoning Processes in Collaborative Visual Analytics," Proceedings of Hawai'i International Conference on System Sciences 44, January 2011, Kauai, Hawai'i.

Arias-Hernandez, R, Kaastra, L.T., and Fisher, B. 2011, "Joint Action Theory and Pair Analytics: In-vivo Studies of Cognition and Social Interaction in Collaborative Visual Analytics." In L. Carlson, C.Hoelscher, and T. Shipley (Eds.), Proceedings of the 33rd Annual Conference of the Cognitive Science Society (pp. 3244-3249). Austin TX: Cognitive Science Society.

Posters:

Arias-Hernandez, R., Kaastra, L.T., and Fisher, B. 2010. "Pair Analytics and Joint Action Theory: A Research Protocol to Study Cognition and Interaction in Visual Analytics." Poster presented at CogSci 2010, Portland, OR. August 12th, 2010.

People:

  • Dr. Richard Arias-Hernández

  • Dr. Brian Fisher

  • Dr. Linda Kaastra