Question-Answering chatbot about initial symptoms
This chatbot contains some training data, the domain file for NLU dialogue in Rasa Core. In the section of actions folder is the main for deverlop, have some data for work control about medical conversation. The THAI MED BOT
consists of the following files:
- actions/actions.py contains actions for control QA Medical conversation.
- actions/semantic_search.py is the semantic search model for use symptoms to tell diseases.
- data contains training data for NLU Dialogue.
- domain.yml contains the domain of the assistant.
- credentials.yml contains credentials for the channels like Meta Messengers.
in initial we have thaitokenizer.txt
file is the manual for install rasa and pythainlp's tokenizer to use (for pythainlp's dependencies we use python3.8 is component)
when you've already installed and followed in the thaitokenizer.txt
. so do the next step:
- Train a Rasa model containing the Rasa NLU and Rasa Core models by running:
rasa train
- Open in the another terminal and open server with Action Server for can use
actions.py
andsemantic_search.py
file
rasa run actions
- Now you can chat with my bot in the following command:
rasa shell
Or if you want to see how Assistant can get slots or how bot can interact each action step use command:
rasa interactive
for use in the Meta Messengers channel, using the following command (but you have to connect with the webhook and open sever like ngrok to get response before using this, see in the Documentation) :
rasa run --credentials credentials.yml
Thanks for how to use pythainlp's dependencies in Rasa : https://rasahq.github.io/rasa-nlu-examples/docs/tokenizer/thai-tokenizer/