Why is this interesting?
· We have automatically tagged the topic of conversation without doing any kind of special training or characterisation of systems to the dataset. We sit back and see what emerges.
· This is a summary of approx 10,000 tweets generated in real time – something that would take many hours of researcher time to put together manually.
· The content of tweets reflects the running order of the programme (see http://www.bbc.co.uk/programmes/b00wmz3q). The topic changes from student tuition fees (represented by the dark green “education” topic and purple “money” topic) to prison policy and then wikileaks (content of tweets shifted to “crime” – light green trace).
See http://www.voiceyourview.com for more project information or follow @voiceyourview for the latest.