Remove dates for books with ISBNs with publish date < 1400

[mekarpeles]

Problem

Some books that have ISBN have very old dates that may suggest a metadata inaccuracy. We should try to find a heuristic to remove dates from items that are clearly wrong (e.g. the date 1 in this set seems suspect)

Relevant URL(s)

https://openlibrary.org/search?q=publish_year%3A%5B*+TO+1400%5D+AND+isbn%3A*

Notes from this Issue's Lead

@judec and @hornc should have more details

Proposal & constraints

NB! Some works were published pre 1400s and then were re-published as modern editions with ISBN so we may want a way to address this.

Related files

Stakeholders

[scottbarnes]

4619 editions with pre-1400 publication years that had their dates removed:
pre-1400-publish-year-with-publisher--date-popped.jsonl.gz.

The criteria used were a publication year < 1400 and the record had a publisher listed. However, no publisher existed then, so these editions likely have incorrect dates. 123 editions were identified as Arabic by having /languages/ara or by the title field had >= 50% off its characters identified as Arabic unicode. These should be examined to see if their dates were likely entered in the format of the Solar Hijri calendar.
arabic_records.jsonl.gz