Wednesday, August 31, 2011

Reaction: The Value of Information Visualization

The paper, entitled “The Value of Information Visualization” emphasizes the difficulties inherent in recognizing the value in information visualization. The paper presents several arguments and examples to help elucidate the value inherent in these techniques, provides background information and scientific views from other fields as validation of visual theoretic principle, gives specific cases where visualization was useful in history, and provides a reckoning of the difference between visual and automatic analysis.

The first argument posits that, assuming information visualization is most useful as an exploratory aid, then we can determine the situations in which information visualization is most powerful and likely to show its use. A second argument identifies the use of information visualization as a tool for accelerating scientific thought through a rather limited quality of human understanding, that is, we are better at pattern recognition than other things. The paper then goes on to analyze qualitative anecdotes for validity by providing specific examples and determining that those examples prove the validity of the argument; a circular and incomplete analysis. The paper proceeds to sum up the work of other authors in the field of enhanced cognition through perception and the relative use of vision as compared to other sense. It then provides several examples of “success stories” in which visualization supposedly had a significant impact on society, a rather dubious claim since it is not possible to determine if it was the data or the packaging of  the data that had the impact.

The most significant aspect of the paper was the comparison of information visualization and automatic analysis of data. In this case, scientifically rigorous experiments were performed to determine the quality of information visualization. The first example of these, which was quite qualitative, was providing statistical data, as well as the data layout for several datasets with similar profiles, but very different in shape. The next used data mining as an example to show, with noisy data, it is often difficult to determine analytically, the profile of data sets if there are different groups of data types mixed together.

Finally, the paper analyzes the economic impact of information visualization, which seems to be the most effective means at determining the usefulness of information visualization. In it, we can determine, not how a visualization works, but it’s overall impact on “profit”, used in the most liberal way. In this case it is possible to measure and compare such things as speed of information retrieval.

Overall, this paper provided some, but not a significant amount, of high quality quantitative data capable of being analyzed. Mostly, this paper vacillated between anecdotes and unverifiable theories of science from often less rigorous and/or analyzable fields. This paper could have provided more information, in perhaps a visual form, showing quantifiably the benefit of information visualization instead of throwing it in towards the end in the underdeveloped section of economic impact.