Automatic Design of Graphical Presentations

Posted on January 1, 2016
Category: bibliography

This [1] and the more available [2] is interesting in a lot of reguards. Most importantly it provides a computational companion to the the graphical procedures that Bertin[3] lays out. This idea of creating correct graphical represenations appears again and again. Tableau's pallette of suggestions was/is a modern implementation of these ideas.

More subtly, there are many small considerations about the best display of information which also need to be resolved in the creation of a visualization. For example, labeling points in a scatterplot has many pitfalls. Labels may overlap, or to avoid overlapping may appear to be related to points which they are not. Cluttered displays may faithfully recreate the underlying "facts" but this is useless if people can't see it.

Fundamentally, graphics are treated as language unto themselves in which visual "sentences" can be expressed. Comparing the visual sentence that a graphic representation expresses to the logical sentence which it is trying to express provides a usefull way to evaluate the graphic. Good graphic representations express only the logical relations in the data that underly them.

[1] J. D. Mackinlay, “Automatic design of graphical presentations,” PhD thesis, Stanford Univ., CA (USA), 1987 [Online]. Available:

[2] J. Mackinlay, “Automating the design of graphical presentations of relational information,” ACM Trans. Graph., vol. 5, no. 2, pp. 110–141, Apr. 1986 [Online]. Available:

[3] J. Bertin, “Semiology of graphics, 1967.” reprinted by University of Wisconsin Press, Madison, WI, 1983.