Monday, 11 October 2010
Tryinig to Connect Google Earth and Second Life
Above a Second Life model of the Völkerschlachtdenkmal war memorial in Germany
Above a Google Earth model of the Völkerschlachtdenkmal war memorial in Germany
As I have blogged about before I am interested in the idea of using the Semantic Web to connect Virtual Reality to the Real World. But this example shows the complexity.
On the surface it might seem simple, the SLURL is a URI. The google maps link is a URI. Call the SLURL SL URI and the Google Maps GM URI. It seems fairly simple to have the following two RDF triplets:
Leipzig Völkerschlachtdenkmal is located at SL URI
Leipzig Völkerschlachtdenkmal is located at GM URI
SL URI is the same location as GM URI
And some obvious logic that if two places have the same location they are both models of the same place.
But is it this simple.
No. Firstly a GM URL is based on a position on the planet. Not only do models have this but so do Tweets, photos posted to Flickr, foursquare checkins, Facebook places, just about any piece of information has this.
Therefore you would also need to explicitly class the SL URI and the GM URI as models. That is you would have to define in your Semantic web that there are models of things.
This would get very confusing when you would try to decide that a model in Second Life was like a model in Google Earth. In many ways they are not. For example Google Earth model is like a real world thing in that there can only be one of them. Google only allows one model of a specific place or thing on Google Earth, while it allows many photos. So it would seem that a model, in Google Earth, has to be unique if the place is unique.
Not the case in Second Life, where many different styles of models exist. Some places like London are modelled in various different degrees from full scale Twinity like models of certain places to Amusement part like highlight models of major buildings in a city put together to artistic models to models that seek to capture the mood but really look nothing like the real thing.
I have visited all kinds in Second Life. So I would have to teach the Semantic web how to distinguish very valid models from less accurate.
And in SL some models are from different times. For example Grenada Spain's Al Hambra is modelled in great accuracy in Second Life, but the model is intended to be about 400 years old. So precisely mapping it to the Google Earth model of the same place is not perfect, because though they are both located in the same place they occupy different times.
For example I have no idea how to use RDF to assert the statements like:
"Bush Gardens in Florida is modelling people imagination about Africa, and though there is nothing African about it millions of people go their feeling they have an African experience"
Which is the difference between a formal mapping technology like Google Earth, which is probably well suited to RDF, vs a experience environment like Second Life, which is open to imagination and fantasy and is much less so suited. I am not even sure how you would make RDF assertions about RDFs own fitness to cover certain domains, and make maps like the one above in a context of Second Life's ambiguity.