Together with the tweet-O-meter project run at CASA as part of the NeISS research project we have collected location tagged tweets around London (M25). As described in an earlier post on this HERE, the idea is to capture the urban narrative. The current data covers a whole weekend from Friday evening to Monday morning and the set holds some 380'000 individual tweets. However this brakes down to 60'000 truly geo referenced tweets, by 5'500 individual users. The thing is, that these are only the mobile tweets and they are captured only if the locations sharing is activated in the twitter profile. Still this makes an average of 10.6 tweets per mobile user over the weekend. Overall we have 39'222 individual users witch makes some 9.7 tweets. So the mobile users seem to message slightly more, but not significantly as one could maybe expect.
In terms of density per location as one could expect the focus is in the centre. There are local hotspots as the weekend progresses, such as Kings Cross and Old Street. But then there seems to be a accumulation of density along the transport lines into and out of the centre.
To visualise the temporality of the data tweets are in the below clip output as a message cloud rising and hovering above London. It is a simple time-space aquarium were the time is plotted as the hight information. The later in the weekend the tweet is sent the higher above the city it floats. As the density develops the low times can be clearly spotted, when it thins out the lines and London sleeps. The animation is rendered in Google Earth, with the KML file brought in through a VB script from Excel. Once set up this is quite a flexible combination. However, the KML file can get quite big, since there is a lot of information contained with all the messages.