Ad hoc trend tracking (or where’s it snowing?)

The heavy snow falls last week in the UK have caused quite a lot of chaos, a lot of which is due to not knowing how bad off different areas were. Ben Marsh created a mashup of a snowmap last Sunday, before snow had fallen over much of the UK, which plotted posts on twitter with the #snowuk tag against a photo and report of the snow level. This shows how quickly it is possible to directly track things over the internet using an ad hoc organised system over services like Twitter.

Contrasting to my last post, where the trend was tracked indirectly, this is directly tracking on outbreak of snow with tags from people. This is tending towards ideas from the semantic web by making information more machine readable, and twitter is encouraging this with the internet population. This is not done in a semantically strict way though but using ad hoc tags and meaning decided by people as they are needed, as we have learnt from the past with the web.  This is probably more the way people would want to interact with the semantic web, but it does remove some of the key ideas of interoperability and data linking.

Mashups like this won’t be able to accurately or reliably predict anything but they do have a very fast reaction time. This makes them very useful to people, which is what matters, so how can we start using systems like this on a more global scale?

Disease trends and blog posts

The project for my M.Sc. this summer is going to be about tracking disease trends through blog posts. This will mostly likely be tracking Influenza as it is the most common disease each year and has readily available data, especially in the US, though I hope to include the UK and other countries if I can too. I’m not sure how reliable or accurate this is going to be, but there’s only one way to find out.

Tracking disease trends through indirect measurements is starting to become quite an interesting area in research. The internet has a vast number of users and small lag time between events and publication which makes projects such as Google Flu possible to accurately predict trends faster than direct means.

New social and conversational websites such as Twitter and Facebook also bring a lot more data directly posted from users in an almost real time basis that could improve disease tracking even more if the data is use in the right way. With website like these still being quite closed with their data for commercial, privacy, and technical issues it is still away off before researchers can go further with this.

I’m going to post my progress on this blog as the project develops to try and be as open with this research as possible with the hope that it can be applied to other sources of data than I have access to.

 

Turn Me Up!

The new Elbow album has “Turn me up” certification.

I’ve not heard of this before but looking on their website they are promoting greater dynamic range in music and stopping the over compressing of music that the music industry do today to sell CD’s. This is great as many CD I’ve listened to over the last few years were quite flat and harsh to listen to. Looking at The Vines album Highly Evolved a few years ago I found that the sound was peaking in most of the tracks which is horrible! Hopefully people will start to listen to music properly once this gets going and more producers start paying attention.

Also on Elbow, their live session with the BBC Concert Orchestra for radio 2 is amazing, if you in the UK you can listen to it here, it’s also available from the usual sources on the web…