The chemist Peter Murray-Rust speaks passionately and powerfully of the ways in which traditional presentations of information in scientific articles, such as graphs and charts, actually obscure or destroy data, invoking scenes of readers employing rulers to try to estimate the actual values of coordinates of points in a graph. 3 Clearly, in a digital environment, it would be much better to be able to move directly and easily between the underlying table of numerical values and their graphical representation, for example. Note such a table really is, in my view, intellectually an integral part of the article rather than underlying data (it may in fact be a complex derivative or reduction of the underlying data, or it might duplicate it). While the example of a two-dimensional graph is fairly straightforward, one can imagine a wide range of more specialized visualizing tools operating on various forms of structured data.
To me, resolving this problem implies a somewhat “thicker” layer of software mediating between the machine-readable representation of articles in the cyberinfrastructure environment and the human reader. Today, articles are most typically delivered to readers in very unexpressive, semantically limited forms such as PDF or HTML, which are rendered by a PDF viewer or a web browser, respectively. As we build out the collaboratories and virtual workspaces to house the activities of our virtual organizations within the cyberinfrastructure, I hope that we will see a new generation of viewing and annotation tools deployed, presumably working on semantically rich XML document representations that will allow us to move beyond the kind of practices that Peter Murray-Rust so appropriately highlights as impeding the progress of scholarship.
The issues here are not limited to graphs and charts. Let me just give two other examples to illustrate the range of potential opportunities.
The astronomer Robert Hanisch gives a talk4 that includes this compelling example: An article includes an image of an astronomical object captured at a specific frequency range; a reader would like to place this object into the context of the available archive of astronomical observations and see what additional observations might be available at other wavelengths, for example. The process of manually re-creating what is, in effect, the digital provenance of the published image proves to be quite arduous, though once the source image in the archive context is established, it’s easy enough to check for the availability of additional imagery for the same region. Clearly, it would be very desirable to generate the trail of digital provenance as the image is prepared for publication, and to make an appropriate representation of that provenance available in the article along with the image to facilitate exactly the kinds of exploration Hanisch describes.
