Collect data to monitor rasa-core and rasa-nlu

Hello everyone,

I was looking for some way to get rasa-core and rasa-nlu data, like confidence, intent, entities, Policy used, etc. I would like to use this data to monitor my bot, for example:

  • What is the most common dialogue path?
  • What is the most matched intent?
  • Which intents are not being asked at all

Does anyone knows if there is any “watcher”/“hook”/“tool” for doing that on rasa-core?

Thanks

I am not sure if every information you mentioned is stored there but A. you can add a tracker store (which by default is InMemory) as MongoTrackerStore or RedisTrackerStore ( see tracker_store.py) and look in the forum for a detailed answer regarding how to use the tracker store. Redis you can add a Pub/Sub mechanism to pull changes to a tracker real time. A tracker is the history of a conversation with a particular user. P.S Redis should be used as a temporary cache. You can also use the MongoTrackerStore as well

B. Not really efficient but you can put debug on, and listen to the detailed logs

1 Like

Hey there — I’m one of the co-founders of Dashbot — we are a full analytics solution, built from the ground-up, specifically for conversational interfaces. We just launched our Rasa integration and would love to see what you think.

Documentation is here: https://www.dashbot.io/docs/rasa/

1 Like