Hi everyone,

I am using iNLP for topic detection and just trained a TA-Model with a csv containing emails and mapped topics.

Now, the users want to train on several ways at a time:

• add Feedback on production-stage on UI (this should work, and just retrain model with using feedback-data)
• Export wrong/not detected cases from production-stage as Excel, add their right topics in the excel and import this on QA-Stage (to test), then after testing, deploy this model to production.

Questions:

1. Is it possible to deploy the exact model of QA-stage on production stage? (How is working the deployment with .jar file?)
2. When I am always training on QA-stage, and deploying on production, do I have the same data basis, or is the model on production "learning himself" and dynamically changing?
3. How do I also use the Feedback Training Data from production for training on QA-stage?

Many Thanks in advance!

Taha