Price Watch | Food Crisis ChatBot

Built with Streamlit and Rasa.

Features

• Analyze food prices across multiple countries on a given time period.
• Display available commodities/countries/periods for comparison for given countries/dates/commodities.
• Automatic translation to any language with Google Translate integration

Requirements

• Python 3.8

Setup

1. Install required Python packages and SpaCy model:

   pip install -r requirements.txt

2. Download SpaCy model for English language:

   python -m spacy download en_core_web_md

3. Download fasttext language detection model (126MB):

   curl -o .\lid.176.bin https://dl.fbaipublicfiles.com/fasttext/supervised-models/lid.176.bin

4. Load HDX datasets and set up actions server:

   UPDATE_HDX_DATASETS=True rasa run actions

5. (optional) Update lookup tables for country names and commodities:

   python datasets/collect_lookup_tables.py

6. Run Rasa training (it will take a few minutes):

   rasa train

7. Load model endpoint:

   rasa run

Interact with chatbot either with shell:

rasa shell

Or with streamlit application:

streamlit run streamlit_app.py