How did Energy of the Nation work?
Every Olympics and Paralympics comes with the highs, lows, bitter disappointment, surprises, triumphs and glory. And London 2012 had it all. In today's digital world these feelings were conveyed within minutes across social media.
We worked with the UK's foremost expert on Twitter sentiment analysis, Professor Mike Thelwall, from the University of Wolverhampton, and Sosolimited (), a team of MIT graduates with expertise in linguistic analysis and data visualisation, to provide a robust methodology, state of the art analytics and accurate results.
Throughout London 2012 we measured the Energy of the Nation by reading the raw Twitter feed and filtering it for those tweets originating from the UK that made reference to the Olympics and Paralympics.*
The content, hashtags and links inside the tweets were scanned for words and phrases like "Olympics", "Paralympics", "Torch Relay", "#energy2012", "London 2012" and related terms to determine their context.
Our library of words was vast and featured traditional phrases, modern terms and even colloquialisms used around the country from Aberdeen to Anglesey.
The location of the tweet author was determined through GPS coordinates, IP address, or self-reported place information.
If tweets featured the words we were monitoring, energy was added to The Energy of the Nation, but don't worry, no data was taken for marketing purposes, we just wanted to give everyone a better view of the way we were all feeling about London 2012 and celebrate it.
Tweets that made it through the filter were passed through a sentiment algorithm that determined the amount of positive and negative emotion contained in the message.
So, emotional words like "brilliant" (+3) and "failure" (-4) were given scores and for each sentence, booster (or negative) words such as "very" (+1), "almost" (-0.6), or "never" (0) we modified the emotional score of any of the words they are found near.
Punctuation was also taken into account to boost (!!!) or reduce (?!?!) the energy count.
A large library of positive and negative emoticons like :( (sad face) or :) (smiley face) were also used to discover 'non-verbal' sentiment.
By tallying these counts of positive and negative scores over time, charts of the daily energy of the nation were made. The ratio of positive to negative energy was then expressed in a simple pie chart.
We integrated the positive and negative charts over the entire day and divide the two values, for example:
This process and the supporting algorithm provided a single, easily understood value that expressed how positive the nation was each day.
The Energy of the Nation!
*This data shows the peak reading of sentiment from Twitter conversations about London 2012 taken daily. See How it Works