One of the most powerful tools we offer is NLU tagging. This is only available through the API, as it is too much information to deliver to the front-end web and mobile app platforms. But for anyone using our API, it can be an essential tool.
Two Types of NLU Tagging
There are two delivery methods for these tags.
The first is tags extracted from stories, which will be returned on the /stories endpoint. Any entities that we recognise from the content we processed will be tagged and provided for the story.
The second type is actually a service for indexing your own internal content. Clients send their textual data to the /entity_extraction endpoint and we will process the data then return the tags we've picked out. We will also return any associated data from our database, too. For example, if we picked out the entity Apple Inc, we will return Apple's ticker, legal IDs (SEC and IRS numbers), and its economic hierarchy, including sector, industry, and category.
What are NLU tags?
NLU tags are identifying tags attached to various words in a sentence. These tags can be
- companies
- events
- locations
- amounts/numbers
- topics
- and many more entities
This sentence can be split up as "company given a price event at rating
agency"
These tags can be very useful when doing automated analysis. You might want only headlines that contain products, or you might be searching for headlines that contain people and amounts to find analyst projections.
As part of our API, these come in a machine-readable format, so you can do almost anything with them, limited only by your programming skills and creativity.
You might even be able to combine tags with different analytics, like Sentiment Analysis, to see if mentioning a person along with a product increases or decreases the Sentiment score.