wrote a blog on an anime recommendation system I built recently.
it discusses how I
- collected the world's largest anime dataset (30.5k rows)
- the architecture of the recommendation engine
- user profile clustering to capture distinct moods
- used dsu to reduce franchise spam