Sprout is an interactive anime recommendation engine. It uses a machine learning model to generate tailored
recommendations based on your watch history and ratings from MyAnimeList or AniList. It also provides a
variety of filters and options for narrowing the results.
In addition to recommendation, Sprout also provides a growing variety of stats and visualizations of both your
personal anime watch history and preferences as well as the entire world of anime via the Atlas Visualization
Sprout is fully Open Source. This includes the full website
code as well as the recommendation model and data collection scripts.
If you encounter any issues or have any other feedback, please feel free to Email Me
or DM me on Twitter @ameobea10. I'm particularly interested in
feedback about the recommendations - please do let me know what you think. I'm interested in making Sprout as
good + useful as it can be!
About the Recommendations
Sprout loads your anime list from your MyAnimeList or AniList profile and uses it to generate a list of
recommendations of animes that you are likely to also like. It takes your ratings into account as well — low
ratings are treated as negatives for the model.
In order to do this, it uses a neural network that is trained using data from other MyAnimeList users. Unlike
some other recommendation strategies that generate recommendations on a rating-by-rating basis, this model has
the advantage of being able to consider your entire anime list at once. This allows it to make use of complex
relationships between anime + ratings to produce higher-quality recommendations.
If you don't have a MyAnimeList or AniList account, you can use the interactive recommender
or check out some
About the Atlas Visualization
The Atlas Visualization is an interactive map of the world of anime.
It is built by using data about relationships between different anime derived from both user ratings as well as
anime metadata to place more closely related anime near each other.
If you load in your anime list to the visualization or view it from your user profile page, it will highlight
anime that you have watched with a blue glow. This is useful as context when exploring the atlas - I highly
suggest doing so!
The atlas visualization was built by constructing a high-dimensional Graph Embedding which encodes the
relationship data between all the animes and then projecting that down into 2D using t-SNE.