Professor Ben Schneiderman speaking at Microsoft Research Cambridge, March 5th, 2009.
Here’s a bit of background: http://www.ukupa.org.uk/cambridge/archives/000627.html
I was lucky to have been quick-off-the-mark, and get my name down to go to this talk, since it sold out very quickly, apparently. I say sold out, but by some miracle of sponsorship or goodwill, it was entirely free! Lucky us. I found it very engaging, and it gave me lots to think about. And I spell visualisation with an “s”, OK? It just doesn’t seem right with a “z”!
I was able to scribble down a few notes and interesting quotes, and if anything, they might be enough to encourage you to go off and read more about the subject. How it all applies to your web site / app / interface / whatever… well, that’s up to you to think about! :)
Resources and Links
Slides, handouts and even video of another version of this talk are now gathered together and linked to on the Cambridge UPA website.
Beyond “user friendliness”
Accessibility is only the first step towards usability – that is the end point we should aspire to, as we move beyond having simply “user friendly” interfaces.
Enable users to find what they need, and to understand what they find. Dynamic, novel visualisation can sometimes give us answers to questions we didn’t even know we had. Bear in mind, though, that just like in any representation of data (maps, stats, etc), we can lie, we can see things that aren’t there, we have to deal with noise, and information is only as good as the data on which it is based.
Also, it is worth remembering that just because the human eye spots an interesting looking pattern, it doesn’t necessarily mean that it IS interesting! We still need to carry out rigourous statistical analyses to confirm the importance of features.
Using vision to think
(We are using vision to think [Stuart Card])
Information visualisation is very different to scientific visualisation, in that it tackles a new set of problems and seeks new ways of seeing things. For example, the social reality of groups, rather than just their geographical distribution.
Key areas of info viz:
BS went on to showcase some of the info visualisation software and techniques that he has been involved in.
Spotfire – a multi-variate visualisation tool. One of its features is that the user can have multiple coordinated views created from their data, and changing something in one view (i.e. selection, restriction, etc) changes it in all the other views.
This provides users with a rapid tool for exploration, and has been used for things like gene location and drug discovery.
BS is a proponent of the “overview; zoom & filter; details on demand” model of info visualisation. Apparently, this model has been adapted, adjusted, contested and borrowed by others over the years, but he is sticking to it.
Essentially, it means that we should show the user all the information first; then allow them to zoom in on interesting bits and filter out things they don’t want; then we should provide details (data, metadata, etc) for specific things on demand. GoogleMaps fits this model, and so does the Radio Times online…
Treemapping is a method for displaying tree-structured data using nested rectangles. I’m not sure that it would be very helpful to represent the DOM tree of this web page as a treemap, but is an example of tree-structured data, with nodes and whatnot.
Treemaps are can be use for “space-constrained visualization of hirearchies“.
A couple of examples included the Hive Group (I thought that was a bit of a sinister name!) iTunes treemap.
For me, the most impressive example was Marcos Weskamp’s NewsMap. Stunning!
TimeSeacher was the main example here. BS demonstrated that this type of visualisation could also be used for a variety information types.
The rise and fall of stock values during a year – the user could see all the peaks and troughs throughout the year (overview), could simple drag a box around some period that looked interesting (zoom), select only the stocks that they were interested in (filter), and then click on a peak, say, to get more information about what was happening (details on demand).
There was also a great example involving a person’s medical history – a tool which is being used in hospitals today, both to monitor a patient, and to try and predict complications from recognised patterns.
This kind of searching for and recognising patterns is something that we humans are very well-equipped to do, and many people (not all, of course) respond better to these visual representations, rather than “flatter” numeric or tabular representations.
For example, if we have a dataset, and we use SQL to select something from it, we need to know exactly what question we are asking. Database GUIs notwithstanding, SQL takes time to perform. Spotting features in something like TimeSearcher (or those TreeMaps) can be *much* faster.
Hierarchical Clustering Explorer gives us a way to visualise the information contained in multidimensional datasets. We can investigate things like the strength of relationships between networks of things (people, corporations, Flickr users, etc).
This was at the end of the talk, so BS might have gone through it quite quickly. Also, it may have been rather more familiar to human-computer interaction types in the audience, but I was just racing to try and get my head around the points.
We looked at aspiring to a “Rank By Feature” framework.
- decomposition of complex problems into multiple simpler problems
- ranking guides discovery
- having systematic startegies
Searching around, I found a paper co-authored by BS, the abstract from which my shed a bit more light on this area:
“Interactive exploration of multidimensional data sets is challenging because: (1) it is difficult to comprehend patterns in more than three dimensions, and (2) current systems often are a patchwork of graphical and statistical methods leaving many researchers uncertain about how to explore their data in an orderly manner. We offer a set of principles and a novel rank-by-feature framework that could enable users to better understand distributions in one (1D) or two dimensions (2D), and then discover relationships, clusters, gaps, outliers, and other features. Users of our framework can view graphical presentations (histograms, boxplots, and scatterplots), and then choose a feature detection criterion to rank 1D or 2D axis-parallel projections. By combining information visualization techniques (overview, coordination, and dynamic query) with summaries and statistical methods users can systematically examine the most important 1D and 2D axis-parallel projections. We summarize our Graphics, Ranking, and Interaction for Discovery (GRID) principles as: (1) study 1D, study 2D, then find features (2) ranking guides insight, statistics confirm. We implemented the rank-by-feature framework in the Hierarchical Clustering Explorer, but the same data exploration principles could enable users to organize their discovery process so as to produce more thorough analyses and extract deeper insights in any multidimensional data application, such as spreadsheets, statistical packages, or information visualization tools.”
A couple of diverting examples to play with (if you’re not still playing with NewsMap!):
Many Eyes (from IBM) – upload and visualise your own data
Viz4All – a list of some of the great information visualisations on the web, grouped by type