Follow the docs to install RasaNLU available here
git clone git@github.com:dtfiedler/simple-rasa-nlu.git
cd simple-rasa-nlu
python -m rasa_nlu.server --path projects/default
Once you have the server running, you can do an example request to the following URL:
http://localhost:5000/parse?q=hello%20there&project=rasaBot
You should see the following response:
{
"entities": [],
"intent": {
"confidence": 1.0,
"name": "greet"
},
"text": "hello there"
}
-
Create your own training data using the site available here
-
Download the file using the "Download" button in top right as <PROJECT_NAME>-trainging-data.json and save it in the data directory
-
Update the config_spacy.json file to:
{ "fixed_model_name": "<PROJECT_NAME>", "pipeline": "spacy_sklearn", "path" : "./projects", "data" : "./data/PROJECT_NAME-training-data.json" }
-
Run the command (this trains Rasa based on your input)
python -m rasa_nlu.train -c config_spacy.json
- This should create a new folder in the projects/default directory called <PROJECT_NAME>
- Rerun your server by running the command:
python -m rasa_nlu.server --path projects/default
- Test one of your example queries by going to
http://localhost:5000/parse?q=<YOUR_SAMPLE_TEXT>&project=<YOUR_PROJECT_NAME>