Hacker News Reader

You would probably be better served starting with an ISBN database and just looking for spans that match book titles. Not sure if there's one you can download for free, but I saw one on sale for $675.

NER will give you a lot of entities that you need to resolve against something to check if they're actually books anyway, depending on volume you may not find a free API tier.

View on HN

I was thinking about querying Google Books API to check if NNPs are book titles/authors, however, as you mentioned, NER produces a lot of entities. Luckily, there is Open Library dump[0] which has around 16 million book titles (IIRC) with some metadata.

[0] https://openlibrary.org/developers/dumps

I've been using Amazon Product API for Booknshelf (www.booknshelf.com) by passing "category => book" in the search to query only on Amazon Books. The book search by title is quite relevant. I thought this might be useful!

Reply on news.ycombinator.com