Blurrt

☰

Has #WhyImVotingUKIP hijacked Blurrt?

May 22, 2014 — by

On the evening of 20th May, UKIP decided to engage the electorate by pushing out a new hashtag #WhyImVotingUKIP in order to encourage voting in the European Election on 22nd May. In the intervening 48 hours, the Twitter community has hijacked the hashtag using sarcasm and irony in a typically British way!


Blurrt is running a campaign collecting tweets around the European Elections and analysing them to get sentiment scores for the political parties. The question is has #WhyImVotingUKIP hijacked our sentiment scoring?

#WhyImVotingUKIP was trending all day yesterday, and is currently running in the tens of thousands and counting! As far as UKIP is concerned it is trending for all the wrong reasons, but early in the day the party leader tweeted:

Fortunately, the above is one of the few tweets that has been collected as part of our European Elections campaign. When setting up the key words we use a boolean filter to make sure we collect tweets in context of the European Elections. In the case of UKIP, our search terms are  ‘UK Independence Party, @UKIP, UKIP, @nigel_farage, Nigel Farage, #UKIP, #farage, farage’ and the boolean filter is ‘European Elections 2014,European Elections,Euro Elections 2014,Euro Elections,EU14,#EuropeanElections2014,#EuropeanElections,#EuroElections2014, #EuroElections,#EU14,#EPelections,#EUelections,elections,poll,vote’

As a result we only pooled 771 tweets with the hashtag #WhyImVotingUKIP. This is good from our point of view because the tweets are highly ironic with many of them scoring a high positive sentiment, and yet the meaning of the tweets are negative overall. This would have skewed our sentiment analysis and made it less accurate.

In Blurrt version 2 we are building our corpus of sarcastic (S) indicators by identifying new linguistic patterns e.g. 1) tweets ending in a conflicting phrase/ adverb/adjective/interjection, 2) positive/ negative sarcasm sandwich (not mutually exclusive), 3) positive – Negative/ Negative – positive (sarcastic indicator word). We have observed that domains exist where sarcasm is used a lot e.g. travel, politics and so Blurrt version 2 will assign campaigns to a type of category which would be more tuned to sarcasm in the future.