The Text Analytics show in Boston last week was a sharp contrast to the one a year ago. The energy at the conference was immeasurably better, and the attendees presented most of the talks, which means we all got to see real world solutions of Text Analytics in action. I was particularly happy with the technical competance of the attendees who all seemed to have real problems that they were trying to solve. Whether this is a general trend or an attribute of who the show marketed to is difficult to say, but it was an active and engaged audience.
Perhaps the most dramatic change in this show from the last one was the total domination of Sentiment as the “must have” feature. Last year, almost nobody cared, and this year it was the topic of at least 50% of the talks. Sentiment has definately moved into the mainstream as a feature that needs to exist in Text Analytics solutions, but the user community is still trying to understand what it’s capable of and what it isn’t. The two things that don’t seem to be common knowledge yet are:
- Automated Sentiment’s business value is on the fringes (really good or really bad news)
- Document Level Sentiment isn’t nearly as valuable as Entity Level Sentiment
As these new users continue to dig into sentiment they’ll figure this out, and it will help sort out the real vendors that really understand sentiment, and those that are smashing the feature into their offering becuase customers are asking about it.
All in all a really interesting show, which I’ll continue to write about the rest of this week, with posts about the changing vendor landscape, and a more in depth looks at Sentiment and where we think its going.