Thomson Reuters, the organization behind the OpenCalais semantic web service, has a double-dose of sentiment analytics in its toolset, addressing both the financial services and public- and investment-relations pros. The first dose is designed to help investor trading through machine reading of Reuters news content; its analysis of sentiment, tuned for that specific content, feeds into real-time trading execution systems, with about 80 percent accuracy.
The second dose takes that and tailors it for the PR audience that wants to graphically see how market sentiment relates to their stock price movements and their missions.
“The pros we work with don’t have to read every tweet or blog post, they can get a read of the overall sentiment analysis and directionally it’s correct,” says <a href="Sentiment Analysis Symposium “>Greg Radner, global head of PR Services at Thomson Reuters. He and Richard Brown, global business manager of machine readable news at the company, are both speaking at next week’s Sentiment Analysis Symposium symposium. “And then if they see some outliers they can quickly dive in and read individual posts or comments.”
Radner says what’s really required for this audience is moving from machine-readable to machine-learnable. “The next thing beyond sentiment analysis is understanding what the nature of those conversations are,” he says. “It’s only so helpful to be able to say something has a positive or negative tone, but that doesn’t itself give insight into the nature of the conversation, into what people are really saying.” Using Crimson Hexagon machine-learnable algorithms, ThomsonReuters’ Thomson ONE Public Relations workflow platform users can identify the conversations underway, aggregate them, and categorize them into different buckets. Users can train the engine with a little manual processing on the front end about how to categorize posts, and then have it happen automatically going forward.
Take, for instance, a PR pro for a pharmaceuticals company who needs to get insight into the online conversations around health care reform; in that context, the comments from small business owners worried about the costs they may have to bear can be placed in their own category. “You can place these comments into such a category on a manual basis, then the machine gives you a green light that it’s ready to take it over and do it on its own,” he says. And maybe what you discover from the insight you gain into the nature of those conversations is that the audience whose fears you were trying to allay about high drug costs didn’t get that message, so that you can take action to change that dynamic. “Hopefully then [that new understanding] will be reflected in the ongoing tracking of insight into the nature of online conversations,” Radner says.
ThomsonReuters PR clients ultimately want three things: “To know what the tonality is, gain insight and then alter their message if needed on a broad base,” not just engage one on one to nip negative tweets in the bud before they become viral. “Moving from machine reading to machine learning provides that real insight” part of the equation.
The product formally launched earlier this year after a six-month test marketing phase. “The ways to get the message out, to target — people in public relations understand that part with press releases and contact databases,” says Radner. “But where they were looking for help and weren’t ready before to pull trigger was on the monitoring and analytics side. â€¦Traditional media monitoring has been done with clipping for decades, of course, but online and social media monitoring is new and the analytical piece of its sentiment or insight into the nature of the conversation is very new.”