Tweeters’ feelings towards main UK political parties revealed ahead of #GE2017

An analysis of the Twittersphere conducted by University of Stirling researchers over the past week has uncovered people’s feelings towards the two main UK political parties.

The data shows the average sentiment over the past week towards the Labour Party has been negative overall by approximately around 4%, while sentiment towards the Conservative Party has been positive overall by around 18%.

Professor Amir Hussain is Director of Stirling's Cognitive Big Data Informatics Research Lab.

On average across the week, more than 52% of the sentiment expressed about Labour on Twitter was negative. This figure was almost unchanged based on Tweets sent today (Wednesday, 7 June).

Over the same time period, 59% of feeling expressed on Twitter about the Conservatives was positive. This figure fell by around 6% today (Wednesday, 7 June).

Political views

Professor Amir Hussain, Director of the Cognitive Big Data Informatics Research Lab at the University of Stirling, said: “We have employed a unique big data analytics approach devised here at Stirling, in order to analyse the sentiment of voters who choose to express their political views on social media. Our technology, called sentic computing, extracts natural language 'concepts' to analyse Twitter sentiments, as opposed to other conventional techniques which pull out 'keywords' and 'bag of words'.

“The sentiment trends we have uncovered in relation to the 2017 UK General Election appear to reflect average national opinion polls which show positive sentiment towards the Conservative Party reducing, and that for the Labour Party increasing, over the past month. Our analysis further shows a temporary surge in negative sentiment towards both parties, coinciding with the tragic terrorist attack in London on Saturday, 3 June.

“It will be interesting to see how these social media interactions translate to actual votes cast. If we do indeed see a correlation, social media sentiment analysis could be another useful indicator for predicting future election outcomes.”

The researchers examined nearly one million Tweets between 1 June and 7 June, 2017.