Information Visualization: Beyond the Horizon [Paperback]

Chen gives a masterful excursion into how information can be visualised. A major aim is that it is presented in a way that a human can see a large mass of data in a meaningful manner. Hence many techniques are shown in numerous colour diagrams that are an indispensible part of the book. Indeed, the book would lack much meaning without those diagrams!

The display ideas have mostly been developed in the last 15 years. In part due to increasing computational power and graphics, that makes such displaying feasible. But another driving force has been the Web. And within this, the Virtual Reality Markup Language. Various proponents, like Blaxxsun, have built VRML worlds in which data can be shown. And in which users can browse. Often in a multiuser mode.

One lesson from the text is that simulated annealing is simply too computationally intensive for deciding how to make a graph with #nodes > 100 or so. It's certainly a nice idea. But sadly only for smaller graphs.

There is an interesting discussion on topic analysis and display. A harder problem than "merely" dealing at the document level. But the results shown seem rather limited. Much more work is needed here.

If you are from physics, you should note that an extensive, protracted example of superstring research was used by Chen. He showed his own research in how key papers could be found via co-citation analysis and graphing. This was to tackle the general problem of trying to find trends and paradigm shifts in scientific research.

I had already decided to grade this a four instead of five, in part because it makes me cranky when world-class authors such as the author of this book neglect other world-class pioneers because of their unwillingness to do a proper search outside their own narrow boundaries. I refer of course to Dick Klavens, Brad Ashford, and Katy Borner, whose Maps of Science are online and spectacular. Even Eugene Garfield, the inventor of citation analysis, gets short shrift.

That aside, the book is an essential reference. While it makes the needed point, that first generation visualization was about showing structure and relationships, and second generation visualization needs to be more dynamic and depict evolutionary and revolutionary changes and mutations (and I would add, provide early warning of anomalies and emergent patterns).

The last chapter, 8, on Detecting Abrupt Changes and Emerging Trends, is very interesting, but heavy on mathematics, and lacking in great detail, which reminds me this is really an overview text, and should be valued in that light. Two examples of fraud detection that I have personally seen as representative of the power of visualization include Dr. Bert Little's discover of $79 million in crop insurance fraud among roughly seven insurance agents and 20+ specific farmers; and the brilliant work of Dr. Simon J. Pak and Dr. John S. Zdanowicz who found $5o billion a year in import-export tax fraud (and Colombian coffee cans marked one pound and weighing 1.5 pounds) through their exploitation of public Department of Commerce databases.

This book has been assigned to our senior working technical person along with three others listed below.
A New Ecology: Systems Perspective, Sven Jorgensen et al (Elsevier, 2007), not on Amazon that I could find
Handbook of Data Visualization (Springer Handbooks of Computational Statistics) (Springer Handbooks of Computational Statistics)
Information Visualization: Beyond the Horizon
Building Trustworthy Semantic Webs

For myself, I put the book down thinking to myself, citation analysis is all well and good, but how do we integrate co-visualization of content, geospatial, money (e.g. "true costs" of each aspect or attribute)?

I continue to admire the work of Peter Morville, such as Ambient Findability: What We Find Changes Who We Become. His name does not appear in the index either. See also: Keeping Abreast of Science and Technology: Technical Intelligence for Business

This is a beautiful book, with many color images. While it includes a large number of excellent figures, and covers a wide range techniques and systems, the text is not, however, a good starting point for a newcomer to the field. (See Card, et al, "Readings in Information Visualization" for a starting point.) This book should be used as a pointer to the literature, which provide the missing details. It covers a wide range techniques, and is worth having.