Sentiment analysis tools: but whose sentiment?
Jinfo Blog

By Nancy Davis Kho

Abstract

The next step along from sentiment analysis looks at the "emotional brain" to find out what the individual is really thinking. However as in the case where the majority of online product reviews are provided by a minority of consumers, how can research really be accurate.

Item

In the past few months we’ve published a number of mini-reviews of sentiment analysis tools, like Finif by Heidi Longaberger and Opfine by Scott Brown. These tools are designed to capture real-time fluctuations in the flood of user-generated content around brands and companies, and are critically useful tools for researchers engaged in brand management or competitive analysis. As the reviews have shown, vendors are making great strides in developing easy to implement solutions around sentiment analysis, which makes these tools accessible to all.

Which means it’s time for  a new competitive edge.

Enter MindSight, from Forbes Consulting. The press I read on this product immediately caught my attention because it goes a step beyond sentiment analysis to uncover “the emotional brain” – that is, what is the consumer really thinking when they are moved to action? MindSight works by using neuroscience to create visual surveys that elicit rapid responses from consumers, free of “editing” by the rational brain.

This “digging deeper” to understand not just the what but the why of consumer sentiment is an innovative next step in contextualising sentiment analysis and making it actionable. And that’s critical given an inherent flaw in sentiment analysis: does everyone’s opinion hold equal weight?

This blog post from earlier this month in the Wall Street Journal’s TechEurope site looks at this problem, mentioning a Forrester Research study that found that “6.2% of the US online population generated 80% of all the impressions about products and services in social networks”. Vendors like PeerIndex, Klout, Kred and Swaylo are trying to put metrics around the opinions by quantifying just who is an expert on just what.

As sentiment analysis tools evolve, it will be interesting to see how more transparency in qualified opinions, true motivations and influential opinon-holders can make the tools ever more precise and accurate for their users.

« Blog