Plus d'informations ici
Information Visualization from Trenches and Labs
Intervenant(s) : Nadia Boukhelifa (Télécom ParisTech)In this talk I will briefly describe my experience in designing and evaluating information visualization tools and techniques, that are either: (1) Specialised: following User-Centered Design (UCD) methodology where visualization techniques are designed for a particular user group having well specified needs and tasks (the Trenches part); and (2) General: techniques that are not focused on a particular user group, e.g. grounded by perceptual studies (and tend to be carried out in labs).
I will go into more detail about my work in two topics: "guided visual exploration" and "uncertainty visualization".
Guided visual exploration
We developed Evolutionary Visual Exploration (EVE) as a framework for guided visual search that combines visual analytics with stochastic optimisation to aid the exploration of multidimensional datasets characterised by a large number of possible views or projections. In the first part of this talk, I will present the results of joint interdisciplinary work in the design, implementation and evaluation of an EVE tool called EvoGraphDice. The need for such a tool rose from our close collaboration with biologists and researchers in agronomy. Our method complements PCA, clustering and regression types of analysis, offering additional features such as interactivity and adaptability.
Information visualization can show not only what we know about the data but also the degree of our confidence in that data. Communicating such information is important for more accurate interpretation of the data and better decision making. In the second part of the talk I will report on my previous and ongoing research in uncertainty visualization where we focus on intuitive ways to communicate qualitative uncertainty to lay audiences. Namely, we investigated the use of sketchiness as both a visual variable and an overall rendering style to convey visual impreciseness that may be associated with data quality. Our results show that sketchiness is a viable alternative to other visual variables to convey qualitative uncertainty and that there is a link between sketchy rendering and increased user engagement.
marc (at) null