Will sentiment analysis break through in 2012 - probably not

Image source

The current estimates are that there are over 500 tools that will listen, monitor, track or analyse your business, product, service, brand, PR, reach, influence or customer digital interactions and deliver a dashboard of “data analysis”. The question is now not should we do it or what to listen for but how to read the analysis and decipher what customers are saying or trying to tell you; but not going as far an assuming customer know what they want!? Hence the interest in “sentiment analysis” which aims to give better output analysis delivering better marketing, detection of opportunities and threats, protection of reputation and brand, and maintain or improve margin.

We know that analyzing natural language is difficult (even without accents) however sarcasm and other forms of derisive language adds additional complexity and we cannot assume that customer statements are true, as we know that  context has a great bearing on meaning.

Given I was working on video phones and natural speech algorithms in 1990/ 1991, these are tricky coding issues. One problem we faced then has gone, which was a zero cost method of collecting large quantities of data, but working out the algorithm and making the code efficient is in a different league. My view is that sentiment analysis tools will continue to evolve but am I expecting a major break through in 2012:  NO.

What is needed is beautiful data representation so we can look at the data and spot the trend.  Tools will get there but the jump is still too big for the level of trust and reliability we demand.