Unlocking Book Recommendations with Vibes: An Inside Look at the Future of Reading
Discover how hardcover is revolutionizing book discovery through its innovative recommendation system called Vibes. In this episode, we explore the technology behind personalized book suggestions, how Vibes creates tailored recommendations, and what’s coming next for this powerful tool.
Key Topics:
- The concept of book Vibes: understanding the fingerprint of books based on reading data
- How Vibes uses a recommendation engine powered by vector databases and machine learning models
- Behind the scenes of generating book vectors and how cosine similarity finds similar reads
- The process of training the model with millions of books and the importance of probabilistic genres
- How user preferences and collaborative ranking refine recommendations
- The role of genre data, axioms, and World types in shaping book suggestions
- Excluding books and authors from recommendations and the impact of user feedback
- Plans for expanding Vibes’ features, including API access and customization options
- Visualizing vectors and genres to better understand book relationships
- Future ideas for recommending based on user book clubs, want-to-read lists, and more
Timestamps:
00:00 - Introduction to hardcover's new recommendation feature, Vibes
02:10 - How hardcover knows the data behind your reading habits
03:39 - Creating an advanced book search system that finds what you don't know
04:36 - Building a recommendation engine to match niche book preferences
05:04 - Upcoming features and API integrations for Vibes
06:30 - Exploring trending and personalized recommendation Vibes
11:27 - How Vibes generates up to 250 book recommendations for your taste
12:23 - The technical backend: vector databases and cosine similarity in recommendations
14:45 - Training a two-tower model for book vectors and genre understanding
16:16 - Incorporating user collaborative data for personalized suggestions
17:02 - Experimenting with signals and signals weights to refine recommendations
18:21 - The role of popularity weights and social signals in suggestion quality
19:50 - Handling classic and crossover genres like Pride and Prejudice
22:23 - Why different genres and classic books show up in recommendations
25:13 - Visualizing vectors and the potential for future genre space exploration
26:46 - How axioms and world types help refine genre and book classification
37:52 - Building the system from scratch: the development process and lessons learned
49:16 - Feedback, tuning, and future improvements for Vibes
58:54 - Making recommendations accessible for all users and future API plans
60:41 - Closing thoughts: the vision for discovering books that match your unique taste
This episode offers a comprehensive look at the technology shaping personalized book discovery. Whether you're a niche reader or just love finding your next great read, learn how hardcover’s Vibes system can help you explore books in ways you never thought possible.
1 Dec 2024
#64 - Building Stats
Hardcover Live co-founders Adam and Ste discuss the development of their project, focusing on user feedback, data management, and visual design to enhance the user experience and functionality.
1 Dec 2024
#63 - 2024 Year in Books
Hardcover Live hosts Adam and Ste discuss their 2024 Year in Books project, reflecting on past achievements, planning improvements, and navigating AI art and copyright complexities.
Reach and audience
Public platform figures. Ratings count people who left a rating, not total listeners.
Spotify
5.0 / 5
1 ratings
Contact Hardcover Live
Guest appearances
Books guests
Based on episode analysis; this does not confirm that the show is currently accepting guests.
Host of Hardcover Live?
Claim your podcast to manage its listing and keep your show details accurate.
Pod Engine is an independent podcast discovery and analytics service and is not affiliated with or endorsed by this podcast. Artwork and show content belong to their owners. Full legal notice.
Explore this show Podcast research with Pod Engine
