If you're just getting started with a virtual assistant you might be able to quickly get inspiration for intents/labels by using unsupervised embedding and dimensionality reduction techniques. This video shows a trick that might help.
If you'd like to play around with the notebook yourself, you can find it here; [ Ссылка ]
This video demonstrates embeddings that are fetched from whatlies;
[ Ссылка ]
This video uses human-learn for interactive selection;
[ Ссылка ]
![](https://i.ytimg.com/vi/YsMoGd7sYMQ/maxresdefault.jpg)