That's an interesting question; this was a small challenge for some reasons.
1. Since this is my first project, I wanted to ship something clean but fast (scope down to one feature).
2. I initially wanted to use some APIs. TLDR: Goodreads closed its API on December 8th, 2020. OpenLibrary was not bad, but the data required to be more consistent. Google Books API was alright, but I would have also curated manually in the end, and sometimes some info could have been better. Maybe it's a personal feeling, but it's hard to have a good books API out there (probably due to Amazon owning Goodreads and making sure it's hard for competitors to have good data as they do).
3. I knew it'd be complex to satisfy everyone with the recommendations, but I can do my best. My best is to ensure that what is recommended is included, even if it's a small amount, and that the recommendation is well rated by the community (e.g., on Goodreads).
4. Then, I realized (again) that we're in the age of AI, and I could leverage that (at least for a 1st release and see how it goes). Based on that, I tried different prompts with many criteria until I found one I liked. By doing this, I could ensure the quality of what was curated.
If it evolves in the future, this will definitely need improvement. But, for a first release, it does the trick!
I agree, good book data is hard to find despite so many big sources out there.
And Amazon API is unavailable (unless you register as an affiliate, but you need a working relevant project to get approved in the first place).
I am working on a project to scrape some regional book stores, and I am finding it really hard to find an API which can reliably give all the data needed for a book. Google Books API so far has been working okay but turns out a single book has so many ISBNs based on which region or publication its been sold and there isn't an API that provides the list of ISBNs for all its releases.
Your comment made me want to explore AI, hopefully I could find a way to use it for my project and make my life simpler.
That's an interesting question; this was a small challenge for some reasons.
1. Since this is my first project, I wanted to ship something clean but fast (scope down to one feature).
2. I initially wanted to use some APIs. TLDR: Goodreads closed its API on December 8th, 2020. OpenLibrary was not bad, but the data required to be more consistent. Google Books API was alright, but I would have also curated manually in the end, and sometimes some info could have been better. Maybe it's a personal feeling, but it's hard to have a good books API out there (probably due to Amazon owning Goodreads and making sure it's hard for competitors to have good data as they do).
3. I knew it'd be complex to satisfy everyone with the recommendations, but I can do my best. My best is to ensure that what is recommended is included, even if it's a small amount, and that the recommendation is well rated by the community (e.g., on Goodreads).
4. Then, I realized (again) that we're in the age of AI, and I could leverage that (at least for a 1st release and see how it goes). Based on that, I tried different prompts with many criteria until I found one I liked. By doing this, I could ensure the quality of what was curated.
If it evolves in the future, this will definitely need improvement. But, for a first release, it does the trick!
I hope it answers your question!