A site bot build using rasa-nlu.
-
if you are using Ubuntu
pip install -r requirement.txt
-
If you are using windows machine
pip install -r requirement.txt
The above command will fail for few packages likemitie
. So you have to install the followingpip install numpy,scipy,sklearn,spacy
If you have error while downloading any of these packages download whl files from http://www.lfd.uci.edu/~gohlke/pythonlibs/ -
You will have two parts in this projects
- AI-Engine
The AI Engine part takes care of Natural language understanding It is built using a open source NLU agent called rasa-nlu
- Bot application
The bot application is a flask application that has a Client(Simple UI chat interface), a backend that fetches event details pydelhi conference website
-
Initially you need to train your bot to do that you need two json files
config
and 'training_data'. -
To train your model you can use either 'spacy' or 'mitie' algorithm. For my application I am using mitie.
If you are using
mitie
you should download a.dat
file from https://github.com/mit-nlp/MITIE/releases/download/v0.4/MITIE-models-v0.2.tar.bz2python -m rasa_nlu.train -c config_mitie.json
The training may take sometime
-
Once the training is done you will have a
models
folder created with a timestamp -
Now you have to host this model for your bot application to use
python -m rasa_nlu.server -c config_mitie.json --server_model_dirs=./model_YYYYMMDD-HHMMSS
-
To setup the knowledge base for the bot run
extract.py
under bot_applicationspython extract.py
You will have a sqliteapp.db
created -
To run the application run
python app.py