internetarchive / openlibrary

One webpage for every book ever published!
https://openlibrary.org
GNU Affero General Public License v3.0
5.25k stars 1.39k forks source link

Change "Thriller" query on homepage #9043

Open seabelis opened 7 months ago

seabelis commented 7 months ago

Problem

The Thriller carousel on the homepage queries specific authors which means the same works always appear, and also non-thrillers appear by those authors.

Evidence / Screenshot

Relevant URL(s)

Reproducing the bug

  1. Go to ...
  2. Do ...

Context

Notes from this Issue's Lead

Proposal & constraints

Change the query to https://archive.org/search?query=openlibrary_work%3A%28*%29+AND+collection%3A%28inlibrary%29+AND+%28lending___available_to_browse%3Atrue+OR+lending___available_to_borrow%3Atrue%29+AND+%28%22thriller%22%29&sort=-downloads.

Related files

Stakeholders

tfmorris commented 7 months ago

Wouldn't it be better to query OpenLibrary than Internet Archive? The first scanned volume that that query returns is included because a user review includes the word "thriller" which seems like a dubious inclusion criterium.

seabelis commented 7 months ago

No objection. I'm ultimately suggesting the authors not be the search term.

cdrini commented 7 months ago

What about this? https://openlibrary.org/search?q=subject%3Athrillers+ratings_count:[1+TO+*]&mode=everything&sort=random.hourly&has_fulltext=true

Note the random sort; currently we sort by recently returned, but we won't be able to do that.

tfmorris commented 7 months ago

It timed out 2/3 times that I tried it, but modulo that, I think it's an improvement.

seabelis commented 7 months ago

Error. Hmm... Sorry. There seems to be a problem with what you were just looking at.

We've noted the error 2024-04-10/152624830521 and will look into it as soon as possible. Head for home?

cdrini commented 7 months ago

Gah sorry folks I made a typo in my url! That's what I get for trying to change the url directly :P I fixed it it should work now.

seabelis commented 7 months ago

Better. Let's do ratings_count:[2 TO *]. I'm pretty sure someone added a bunch of 1 star ratings to lots of items right when we launched that feature.

seabelis commented 3 months ago
Screenshot 2024-08-19 at 18 37 47

Four out of six of these titles are not thrillers.