If I had tech skills of any sort, I would just go ahead and do this rather than blare it out into the blogosphere for anyone to have, but sharing is caring. So here is my rather simplistic request for a filter/mashup/visualization.
I want to be able to do an emotive analysis geographically using Google Earth. What I mean by an emotive analysis is that I want to visualize sentiment across regions based on a series of criteria (and truly the more that can be included, the better) such as:
- News stories/cycles
- Public discussion boards/chats
- SMS
- Blogs
I have seen some small-scale versions of this done, most notably (or at least most aesthetically) BBC’s White Spectrum, which took responses to specific hot-button questions and mapped those emotively. It is aesthetically luring and disconcerting at the same time.
BBC White Spectrum. Visualizing emotion for controversial issues.
This is a good approach to analyzing emotive data online, but it doesn’t scale out very well. It required a precise data collection tool (presumably a survey question) and a precise structure for extracting emotional vocabulary (upset, angry, frustrated, glad, happy) and mapping that to larger emotional categories (happiness, sadness, etc.). I am looking at you, taxonomists of the world. This thing would need a fairly sturdy, scalable structure.
More importantly, it would need to collect a lot of non-emotional data, namely location (fairly precisely, too). The filters that White Spectrum have in place are good and map, I imagine, directly to their collected data. But I want this thing to scale out globally.
Filter this.
Now imagine a layer on Google Earth for all of this and an emotive search engine running underneath it. One that trolls the aforementioned sources and pulls emotive vocabulary, geolocations, and contextual data (what they were talking about that caused this emotional reaction). Gathering that data, we can then map it onto Google Earth and monitor emotional activity worldwide. What practical application does this have?
I can imagine some fairly rudimentary things happening on certain fronts. First, it would help analyze the effects of larger social issues on the emotive health of a region, like natural disasters, economic volatility, even political or social unrest. More importantly, it might prove an effective tool for predicting social unrest. We monitor the emotional health of the region much as we would climate conditions or economic activity.
Either way, it would just be plain neat. And when was that not a good enough reason to invent something. So, I throw down the gauntlet. Invent it, you dataminers and creative types. And if this already exists, color me embarrassed, but please do let me know.
Not a lot of emotive unrest here. The view from my apartment in Princeton georeferenced on Google Earth.