Hallo, I decided to build a book recommender from a publicly acquired dataset consisting of 6 million ratings for 10 thousand books.
I built a data explorer app: HTTPS://HyBoRiSdata.streamlit.app
A Recommender Algorithm Evaluator: HTTPS://HyBoRiS-RAE.streamlit.app
And a book recommender based on the Cosine Similarity algorithm and the ratings of the top 1,000 raters: HTTPS://HyBoRiS.streamlit.app
I take no responsibility for anything but if you choose one or two of your favourite authors, rate their books and press “Get Recommendations!” it is surprisingly good.
If you’d rather use Genres to get recommendations you can do that too, or a mix of the two, hence the “Hybrid”, but it takes more ratings to get good recommendations that way. This is the first time I ever conceived, built and deployed a machine learning system, and I couldn’t have done it (at least,not nearly as easily!) without Streamlit, so thanks!
If anyone is interested to see the code it’s at HTTPS://github.com/tom-blanchfield/BoRiS