Our Natural Language Processing (NLP) understanding is almost always right. But now and then, due to nuances and complexities in the English language, the results you see may not always be.
So, we added the ability to easily fix anything our NLP engine didn’t get right. If you see something wrong in a chart you can simply “Right-Click” on the Insight and make it right.
In the example below, the Workbench classified “kill Osama” as a negative associated with Obama. In most cases, “killing” would be considered negative. However, in this specific case, killing the terrorist behind 911, is a positive thing.
Here is how it works: from any chart in the Insight Workbench you can Right-Click and …
- Change Sentiment (Positive, Negative and Neutral)
- Hide from Chart
- Exclude from Topic
This new feature should give you more control of the results and get it just right.
We hope you enjoy using the feature.
Please keep the feedback coming!
Jens Tellefsen, Vice President of Products