I am actually working on something similar with Shepherd.com :)
I ask authors to recommend 5 books around a topic, theme, or mood. For example, here is one that Steven Pinker did on Rationality: https://shepherd.com/best-books/rationality
Then I use NLP to identify Wikipedia topics around the book. And, use that to connect it to topics in order to make it a fun process to find your next book online. I am hoping to use Wikidata to build some really cool connections as we get going.
What I also get from this is how books are connected to each other using human data. So that author 1 thinks that book 1 is similar in some way to 2, 3, 4, and 5. I use those inputs to power the new recommendation system I am launching next month. And, a few months after that a "books like" feature that will use that same data. And, then toward Winter I'll integrate book genre data into that.
I am bootstrapped so we are rate limited on dev time :)
In early 2023 I will add user accounts and a custom book recommendation newsletter based on what readers like in terms of genres, topics, books, and authors. And, once I have that user account I will slowly be moving to collecting readers 3 favorite books of the year. And, then using that data to identify your "Book DNA" and match you to other readers.
Slowly but surely :)
Email me at ben@shepherd.com if you would like to chat or have any ideas.
Comments
If anyone has any good recommendations beyond the ones the model gave me, lmk!
I love what you are doing here!
I am actually working on something similar with Shepherd.com :)
I ask authors to recommend 5 books around a topic, theme, or mood. For example, here is one that Steven Pinker did on Rationality: https://shepherd.com/best-books/rationality
Then I use NLP to identify Wikipedia topics around the book. And, use that to connect it to topics in order to make it a fun process to find your next book online. I am hoping to use Wikidata to build some really cool connections as we get going.
What I also get from this is how books are connected to each other using human data. So that author 1 thinks that book 1 is similar in some way to 2, 3, 4, and 5. I use those inputs to power the new recommendation system I am launching next month. And, a few months after that a "books like" feature that will use that same data. And, then toward Winter I'll integrate book genre data into that.
I am bootstrapped so we are rate limited on dev time :)
In early 2023 I will add user accounts and a custom book recommendation newsletter based on what readers like in terms of genres, topics, books, and authors. And, once I have that user account I will slowly be moving to collecting readers 3 favorite books of the year. And, then using that data to identify your "Book DNA" and match you to other readers.
Slowly but surely :)
Email me at ben@shepherd.com if you would like to chat or have any ideas.