Thursday, February 19, 2009

GIS For Visualization Of Library Usage Data And Holdings

Visualisation of Library Data / Informationszentrum Chemie Biologie Pharmazie / Eidgenössische Technische Hochschule Zürich (ETH)
The Chemistry Biology Pharmacy Information Center of the Federal Institute of Technology (Zurich, Switzerland) has mashed a Geographic Information System (GIS) with library usage and holdings data to create Several Most Impressive Visualizations



System requirements for data visualisation: Java browser plugin, Java Run Time Environment version 1.4 or later. A large amount of data is displayed in a Java applet of 1200 pixel x 800 pixel size and a wide screen is recommended.



Martin Brändle / The DeveloperOfTheProject / Has Requested Feedback / His E-Mail Is / /


ekg said...

This is very intersting, and very innovative. Great job.

mbraendle said...

Thanks Gerry for putting this info on your blog. Since then, our server is overwhelmed with requests ...

I would like to provide some details:
- I had the idea since a long time to provide the user a means for finding the right shelf (I'm pretty sure that I can find notes back until 2004 or 2005). First I thought of using SVG (scalable vector graphics), but then I found by hazard a publicly available graphics library that allows rapid prototyping and actually is often used in arts. The idea changed to visualising statistical library data because we are at the beginning of the year ...
- if you look at the HTML source code of one of the pages, you will see that the Java applet is controlled by several parameters that define the files or URLs that are loaded by the applet and contain either the shelf coordinates and info, the shelf row labels, labels for general objects such as offices, study desks, copiers, etc., and the data that is visualized. With the exception of the general floor plan (which is a plain GIF, JPG or PNG image), all data is in XML format (either files or XML streams from our library system). This means that any library floor plan and shelf data can be visualised. However, there is one restriction: orientation of the shelves has to be horizontal or vertical (up to now).
- the visualisations use an "exploded" view in order to show all the shelf boards: The shelf boards are shifted a little with respect to each other, the lowest boards being at the most outward position of a row, the highest at the inner position. Their lengths are in real scale, but depths were reduced. It is also possible to draw simply a shelf from top, with each board one under the other, and using the real scales. I'm thinking of coloring the shelves according to the colors we use in our star tree.
-if you want to see how the library looks like, visit our website and click on one of the pictures.
-some may ask how the shelf occupation is determined. Actually, this isn't just the number of books, it is the real space the books are occupying (displayed as percent in the visualisation, but internally, we use centimeters). I use an estimation equation that is accurate to about +-3 percent and that I implemented in our library system a few years ago. I use it on a regular basis when I assign the books to our systematic shelfing classification, and I developed the equation because I'm too lazy to get up every time in order to see if there is enough space on the shelf. It also helps to eliminate older works if you run out of space. If you have such a formula, you can also calculate the total occupation of the library floors. For G 5, it is at present 75%.

Best regards,

Martin Brändle

mbraendle said...

There is an update to the visualizations: Dragging the sliders (the grey triangles) in the legend allows to confine the data range. Caveat: In the case of the gradient visualization (year distribution), update is slow because about 100-200K datapoints have to be redrawn.