The most popular current Web 2.0 representation is geography: putting everything on a map. It's a powerful thing to do: when we can SEE how close registered sex offenders are to schools and day cares, we have certain reactions about where our psychic sense of "too near" or "too far" meets the legal/phyiscal interpretation of "appropriate distance." A little bit of information, as has been said many times, can be a dangerous thing. This particular offender/schools mash up does not provide a brushing interface that, say, relates re-offender statistics based on various distances from schools to help confirm whether our sense of dread is well-founded our not.
It is with this caveat in mind that, our group has been thinking about how adding not just mapping but temporal mapping might be for a project we have called musicSpace to integrate a variety of musicology sources for easy exploration. More recently in a project called continuum we'd been looking at how to map rich data sets like classical music onto timelines so that the visualization doesn't implode. That is, if there's lots of stuff going on at the same time in a time line, all the info looks like a big blob, or if you zoom out, you lose the surrounding context. Our challenge was to solve the "too much info=blob; too little=not enough information" dilemma. Inspired by that work, we'd like to take what we learned there and think map thoughts.
Mapping Time
What we are calling Temporal Mapping is not unknown but it's not common. to be clear, temporal mapping has one meaning in discussions of disease tracking for instance that doesn't involve visualizations; spatio-temporal mapping has another meaning in computing. The kind of temporal mapping we're considering is more akin to an example from the Land Cover Institute which on a map over a relatively stable geography shows how population density has grown and spread over 200 years. Other work shows how the geography of a place itself (such as a river valley) changes over time.
'Istanbul was Constantinople now its Istanbul" - They Might Be Giants
Our sense of temporal mapping it turns out is more complex than these example because it turns out we are looking at a variety or parameters that change: in terms of locations, borders change; names change and even the geography can change. One way to reflect this change is to use maps that can present borders/locations that are accurate for a given period - this assumes that various places recognize the same borders/place names. Consider the mapping of Taiwan as a political representation issue. To use the music examples, if a composer created something in the 1700s, the borders of the domains were different and the place names may be too, so we need to have maps with borders and place names that are accurate for that time. As we discuss below, there are other issues that come into play when, to coin a phrase, wanting to co-map points that cross times, and thus cross representations of locations.
Even if we don't want to co-map, but restrict ourselves to single maps, there are some challenges: if we know where a piece was composed we map that; if we know where a composer was born we map that, if we know where a composer first performed a piece we map that.
There are a few data subtleties there: do all objects in a classical music repository now need Lat/Long data associated with them, as well as a date? Even there the temporal bit is not so obvious: there are kinds of dates and kinds of locations: how tease these out so they are clear in the UI? so it's clear a person is choosing to see performance dates/locations rather than composition dates/locations. What happens if a work was known to have been taken out and put away over a range of places and times? How is that stored in an object in order to be represented?
If we put aside that question of the back end data representations and UI finesse for the moment, let's assume whatever it is we want to map in music we can map, the more glaring, basic challenges are how both borders and place names have changed not just over the centuries but even within decades. Maps that only map against geography lat/long have it somewhat easier than mapping against historically/politically accurate representations.
And as always, the question of how to represent the information is non-obvious. For instance, how handle multiple names or boundaries for a place? Only show the appropriate name for the specific time? Show all versions to provide context not only of place but between times? These kinds of decisions become critical when crossing domain representations. For example, what happens when looking for a location in europe that produced the most major compositions of the Romantic Era relative to location(s) in Europe of most significant performances in early 20thC. The borders and place names from the 1700s and indeed even between 1914 and 1920 change several times.
So Temporal Mapping is?
Perhaps a fast way to begin to think about temporal mapping in arts and humanities data that involves people, places and times is to be able to accurately reflect these places as they were interpreted both in their times, and in ours, and to be able view these comparisons from any variety of perspectives - comparatively, relatively.
Animation and Insight
A potential benefit of developing temporal mapping approaches for arts/humanities data is in meaning that is communicated through animation: if we can step through the various places by time of where Beethoven worked - see who else was in the neighborhoods at various points, and correlate that with specific works, and perhaps specific historical events and their key locations, can we begin, almost at a glance, to get a new appreciation of a domain space? Do seeing these patterns animated over time and space and politics and whatever else let us ask new kinds of questions - questions that would have been potentially intractable to ask before?
These are early days for our investigations, but from early scenarios domain experts have given us, the ability to step through time, and to see events of interest comparatively across time and space, is a thing devoutly to be wished. These representational desires are driving our current UI research efforts.