July 2013

Reputation, Big Data and Emotion

Reputation, Big Data and Emotion

 

Extracts of this article appeared on the Guardian Datastore.  A full version of our interview is reproduced below.

"We use external digital content to assess reputation for brands and organisations – assessing people&;s perceptions as revealed through blogs, news reports, social media and forums.  

"Elements of this requires some form of sentiment analysis.  However, we have found the traditional division of positive and negative sentiment fairly obsolete for our needs.

"Not only is it of limited accuracy – there is too high an incidence of false results – but also traditional sentiment analysis fails to relate the intensity of the reaction and therefore leaves you thinking “so what?”

"Instead we prefer to look at emotional sentiment for our Reputation Tracker tool.  This is a better way to understand the ‘heat’ of an issue and make predictions about how people will behave as a result. 

"It also allows you to look for less ambiguous terms; the more extreme the language the less likely the words are being used out of context.

"Take two examples that we can all agree are bad – a train being cancelled because of heavy snow or a rise in the cost of our monthly rail season ticket. 

"Both are likely to score highly for negative sentiment, but which is most likely arouse feelings of disgust and contempt targeted at the rail companies and therefore more likely to change how we travel to work in the long-term?

"Emotional sentiment provides fascinating insight into what we really care about.  

"For example, from emotional analysis we know that the price of our fuel bills arouses 4 times more disgust than the issue of human trafficking.

"Clearly it is trafficking that creates the greater harm and causes most human suffering, but it is often incidents of the more minor, personal injustices that shake us out of our apathy and make us take action.

"For energy companies tracking disgust against brands and pricing allows us to better rank the emotions of customers towards each provider. This was perfectly illustrated by one brand which, when caught up in a pricing scandal, went from being the best to second worst for its association with disgust.   "Tracking how such extreme emotion correlates to customers switching to alternative providers, also helps brands to understand their &;reputation elasticity&; - just how much &;bad reputation&; customers are willing to take before they take their business elsewhere.     "Or conversely, how improving reputation would make them more likely to retain and attract customers - a valuable tool in proving the financial value of being a better business."