Launching 🔎 CTRL+F (NLP API to find and replace named entities)
When I was working on another project (Fitmeal), back in 2016, I implemented a large NLP module to find food entities in text messages.
This week, I refactored that code and made it accessible through an API.
$ curl -X GET https://ctrlf-prod.herokuapp.com/food \
-H 'Authorization: Bearer …' \
-H 'Content-Type: application/json' \
-d '{"text": "This morning, I ate an apple with some peanut butter."}'
{"matches": ["apple", "peanut butter"]}
You can find more information about it on the IH Product page or directly on ctrlf.co.
@gduverger thanks for sharing this -- do you have a use case you can explain for this?
How were you using it in text messages and what did you do once you identified those foods?
Cool idea! Are you using Dialogflow ? What are the intended use cases?
Thanks, Levi. I'm not using Dialogflow. At the time I implemented the original code, in 2016, existing solutions were not as developed as they are now. The main use case is for startups or product engineers building a health and fitness app who want to “understand” their users' generated content better.