Information

Please consider supporting this website on Patreon

What is it?

A book like foo is an easy-to-use book recommendation engine that learns and improves over time. You can simply input a couple of books (up to five!) and it will find the best recommendations based on the unique blend of books you've selected.

The system builds better recommendations over time based on the collective queries of all users. It is backed by a database of over 100,000 titles, spanning all genres, and gleaned from a diversity of datasets like WikiData.

How does it work?

Queries are conducted using the following characteristics:

Content heuristics: Genre, Description, Keywords, Author, etc.
Co-associations between books based on aggregated data collection.
Some hand-crafted associations between books/genres/keywords.

Why does it need at least 2 book titles?

In order to give good recommendations we need more signal than that which can be gleaned from a single title. For efficiency's sake, it's best to require people to add at least two books. Otherwise the search space and room for error is too large, resulting in poor recommendations.

Some tips

Prefer querying with a popular book you've enjoyed plus one or two less popular ones. This tends to give good results.
Querying just two books will usually give satisfying results, but will usually only return books that are thematically and predictably similar. Sometimes this is sufficient. Other times it'll help to add another book though.
Add both fiction and non-fiction books to get an eclectic blend. Querying with just fiction will yield only fictional results, and likewise with non-fiction.
Querying Harry Potter books only will probably not get you very far... Try less popular books.
Recommendations are hard-coded to be slightly biased against popularity but you should expect that some very popular books will still be highly ranked (they're typically popular for a reason).

Originally engineered by James Padolsey.