This is a Virtual Assistant Kit which predicts and answers FAQ.
This kit aids rapid development of Virtual Agents by following below steps.
1. Select a development environment of your choice
2. Explore and analyse the dataset
3. Cleanse and get the noise-free data
4. Compute embeddings for the dataset - sentence or word embeddings
5. Preprocess the user query
6. Compute embeddings for user query
7. Compare and compute similarity score to find a best match
8. Look up the dataset for displaying answer of a best matched query
9. Precomputed embeddings can be persisted for later use
10. Servers and webframeworks can be leveraged for servicing the request as REST API
You can also find github reference to the Virtual Agent repo using this kit at the bottom for building your own Virtual Agents.
Jupyter Notebook was used for development and debugging this kit.