This interesting and well-written paper discusses an approach to creating useful, simple visual images based on more complex computer-generated representations. The authors examine two genres: maps intended for tourists, and exploded design assemblies. The work uses human-generated drawings as fundamental templates. As the authors point out, the explosion of complex representations makes it imperative to be able to easily create visual representations for human consumption. The paper describes three design principles: identification, instantiation, and evaluation.

To gain insight into identification, the authors examine what they determine to be useful hand-drawn samples. Perhaps one of the most iconic is the 1933 tourist map of the London Underground. It is an abstract of the city map, showing the stations one by one as well as the various transfer points. As millions of city dwellers and tourists can testify, it is an elegant solution, its form copied in city after city.

The authors then abstract the principles to the problem of generating route maps or tourist guides. They develop a set of attributes, such as position, size, and the orientation of roads and landmarks, to instantiate the map in question. The simplified map provides only information of interest to the targeted audience--in this case, tourists. One example uses what is likely a Google route map showing the highlighted route in all of its cartographic glory. The hand-drawn route map focuses on turns more than distances and directions. The final visual map is a simplified representation of the actual route.

The authors apply the same techniques to assembly drawings. Starting with the usual explosion isometric drawing, they show a hand-drawn assembly instruction and, finally, a computer-generated simple set of steps to assemble the object.

For all of the techniques, the authors employ user feedback and user studies in an attempt to evaluate their process.

This is an interesting approach to the problem. By necessity, the paper is narrow in scope in order to allow the authors to explore the subject. The paper is entirely devoid of mathematics and theorems, and is easy to read. It also includes an extensive set of references. The authors would welcome “more sophisticated evaluation methodology [to] provide stronger evidence for these models [and advance the work].”