It does according to a post on Vigetengage by former staffer Jen Krupey. In a section of her post entitled “Sentiment Stinks,” Jen wrote that, “Overarching consensus was that no tool does a ‘great’ job getting sentiment right. Human review is still needed.
But that was before Jen or anyone else had seen our ConsumerBase product, which addresses the exact problem she’s highlighting. She’s saying that sentiment analysis stinks because the accuracy/precision levels of current tools are so low that a human analyst needs to clean up the tool’s output to make it usable. That means wasted time, and time is money, so it really means wasted money.
There’s no denying that Natural Language Processing is a hard problem for sentiment, but I think users will find ConsumerBase does a much better job at it and results will require far less manual clean-up.
Here’s an example: For a search on Listerine, ConsumerBase can pick out all these negatives:
That’s great precision, which will save you a lot of time. But you’re still going to want to do some “laddering up” of these insights into “themes” that make sense to the business. For example, you would probably group these all under the theme “Too Harsh.”
We recognize this “laddering up” is essential to doing analytical work, so we’ve baked theming right into ConsumerBase. In fact, once you’ve grouped your insights into themes relevant to the way you look at the world, you can then automatically build a pie chart from the data. Our improved precision and the theming capability should be a big time saver for any analyst whose job it is to track sentiment.