Rory Cellan-Jones on the BBC has been writing recently about a number of social media sentiment-analysis tools being showcased during this British general election period. It’s an interesting topic, not just for the politicians, but for anyone who wants to dig into the soup of online debate to determine what real people think about real issues – or brands.
claim to be able to determine sentiment from textual analysis. They are not the only ones, with even the social networks themselves springing various analysis tools on the users to determine how they feel at any given second about the political campaigns. Facebook has even designed a dial users can click on to indicate their sense of positivity.
But tools like the positive/negative feedback dials are only useful in the midst of debates, and feedback is generally coloured by what people already think. It’s easy to understand that – just look at Twitter or any major news source today. Each political party is claiming that their candidate ‘won’ the debate, with full analysis of exactly why and how they were the victor.
So tools like Lexalytics are a lot more useful than reaction dials or buttons. To gauge true sentiment, a large volume of data needs to be analysed over a time series without the contributors necessarily being aware that they are being analysed. The moment you know your contribution is making an actual difference is the moment you can start behaving tactically and skewing the result.
Rather like voting. Though millions believe that doesn’t make any difference whatsoever.