The Model is Naive currently and picks on keywords and its synonyms. I have plans to include Reddit data to retrain the models.

Have touched upon approach here- https://medium.com/@19aditiyadav/how-i-built-bookclub-ai-ai-...

Thanks for your feedbacks and these queries X Results would definitely act as sniff test going forward.