A Linguistic Approach to Categorical Color Assignment for Data Visualization.

When data categories have strong color associations, it is useful to use these semantically meaningful concept-color associations in data visualizations. In this paper, we explore how linguistic information about the terms defining the data can be used to generate semantically meaningful colors. To do this effectively, we need first to establish that a term has a strong semantic color association, then discover which color or colors express it. Using co-occurrence measures of color name frequencies from Google n-grams, we define a measure for colorability that describes how strongly associated a given term is to any of a set of basic color terms. We then show how this colorability score can be used with additional semantic analysis to rank and retrieve a representative color from Google Images. Alternatively, we use symbolic relationships defined by WordNet to select identity colors for categories such as countries or brands. To create visually distinct color palettes, we use k-means clustering to create visually distinct sets, iteratively reassigning terms with multiple basic color associations as needed. This can be additionally constrained to use colors only in a predefined palette.

The effect of colour and transparency on the perception of overlaid grids.

Overlaid reference elements need to be sufficiently visible to effectively relate to the underlying information, but not so obtrusive that they clutter the presentation. We seek to create guidelines for presenting such structures through experimental studies to define boundary conditions for visual intrusiveness. We base our work on the practice of designers, who use transparency to integrate overlaid grids with their underlying imagery. Previous work discovered a useful range of alpha values for black or white grids overlayed on scatterplot images rendered in shades of gray over gray backgrounds of different lightness values. This work compares black grids to blue and red ones on different image types of scatterplots and maps. We expected that the coloured grids over grayscale images would be more visually salient than black ones, resulting in lower alpha values. Instead, we found that there was no significant difference between the boundaries set for red and black grids, but that the boundaries for blue grids were set consistently higher (more opaque). As in our previous study, alpha values are affected by image density rather than image type, and are consistently lower than many default settings. These results have implications for the design of subtle reference structures.

Whisper, don't scream: grids and transparency.

Visual elements such as grids, labels, and contour lines act as reference structures that support the primary information being presented. Such structures need to be usefully visible, but not so obtrusive that they clutter the presentation. Visual designers know how to carefully manage transparency and layering in an image to balance these elements. We want the presentation of these structures in complex, dynamic, computer-generated visualizations to reflect the same subtlety and comfort of good design. Our goal is to determine the physical, perceptual, and cognitive characteristics of such structures in a way that enables automatic presentation. Our approach to this problem does not try to characterize "ideal" or "best," but instead seeks boundary conditions that define a range of visible yet subtle legibility. All presentations that are clearly bad lie outside of this range, and can easily be avoided. In this paper, we report three experiments investigating the effects of grid color and spacing on these boundary conditions, defined by manipulating the transparency (alpha) of thin rectangular grids over scatter plots. Our results show that while there is some variation due to user preference and image properties, bounding alpha allows us to reliably predict a range of usable yet unobtrusive grids over a wide variety of conditions.