There aren’t tech companies any more, there are just companies. That truth is fundamental in the future of Salience. Eventually, the only requirement to run a text analytics job will be a data set—no degrees, no coding experience, no robust data science budget. Just enough money for to run the task and a couple of minutes to spare.
With Salience 6.2 we’ve reinforced and built out a number of features, giving everyday business owners more control over what is being said about them across the internet—from quantitative or qualitative surveys to emails, Yelp reviews to Tweets, and much more. The idea text analytics should be intuitive, comprehensive and nimble.
Easy tuning is key to effective text analytics, “With Salience 6.2, we are dramatically bolstering our ML capabilities, making it easier for our customers to teach and ‘tune’ the software to meet their unique needs,” said Jeff Catlin, our CEO. This was the impetus behind our new, easy-to-use HSDTrainer. With it, you can guarantee your analysis will be perfectly tuned to your requirements.
This corrects a common problem endemic to text analytics; often times the machine will be fed a phrase or entity that is mentioned negatively throughout a data set. The word or phrase might be positive or neutral. But, because it’s been mentioned negatively so many times, the machine is tricked into believing it is itself negative. The HSDTrainer lets you see this inaccuracy immediately. You can simply retune it to the desired sentiment, giving your analysis consistent precision.
Expanding the reach of Salience hit a milestone with the release of 6.2, as it can now strip and ingest email databases—eliminating headers, footers and duplicate emails allowing for efficient analyses. Say you’re aiming to gain actionable insights from customer support emails. Email processing can help you detect silo-ization or improve your ability to react nimbly to recurrent customer feedback that might otherwise fall through the cracks.
Emojis & More
All of this is helped by emoji analytics. Emoji analytics are are a standard out-of-the-box offering. It gives you minable insights from this universal language. Along with emoji analytics, we also improved our precision and recall scores, known as F1 scores, by up to 25 percent. This means improved Named Entity Recognition (entities are people, places and things), guaranteeing you even more precise results. All this leads to the goal set out for us by Jeff: “[Improve] the way machines and humans interact by expanding the capabilities of what text analytics and NLP can accomplish.”
Want to learn more? Check out our website!