Incorrect automatically derived dates for works

DiscussãoBug Collectors

Entre no LibraryThing para poder publicar.

Incorrect automatically derived dates for works

Dez 19, 2023, 9:20 pm

I was surprised to be informed that the earliest publication date for a work I added in 2023 was for The Chief: Biography of Field Marshal Lord Wavell by Ronald Lewin with a supposed publication date of 1600. In my catalogue the book was published in 1980.

The author was born in 1914, and died in 1984, but apparently, according to LT, he also published:
Slim: The Standardbearer : A Biography of Field-Marshal the Viscount Slim in 1545,
Churchill as warlord in 1636,
and The Life and Death of the Afrika Korps in 1905.

I thought automatically derived work dates that fall outside the author's life were now supposed to be ignored, so where do these spurious dates come from?

Dez 20, 2023, 2:56 am

I noticed this the same way, an outlier in my annual review that I’ve now corrected by adding a date in CK. To see how widespread it is I looked through the 66 books in my library with original publication dates 1600-1799: roughly half of them are really 17th or 18th century, the rest seem to be mostly poetry collections from the 1980s, together with a few recent books about 17th or 18th century literature.

Mostly fairly obscure writers where it’s not surprising that it’s hard to guess a date, but there was also Christopher Isherwood’s Exhumations where the author’s dates should be well known enough to block that kind of wild error.

Editado: Dez 20, 2023, 5:19 am

The Canterbury tales appears in my list for “1-199”, which seems bizarrely far off for such a well known and documented work. I wonder if that’s because someone filled in “14th century” for one of the CK entries and it got parsed as “14CE”?

Alcestis and other plays is assigned to the 5th century CE instead of BCE. That looks like another problem with parsing CK data.

Dez 20, 2023, 10:29 am

Here's a book shat shows a calculated original publication date of 1605:

All of the 9 copies have only 1980 in the publication date and publication fields. 1605 doesn't appear in any of the book details records. I wonder where it's coming from.

Dez 20, 2023, 10:44 am

Earliest publication date: The Planet Construction Kit by Mark Rosenfelder (1612)

I added CK to the work, so that should change. But I don't see where 1612 came from, since the dates are all in the 2000s.

Dez 20, 2023, 11:42 am

Here's what LT staff have said previously about the dates and where they come from:

2020 (when the series feature was updated): "The publication date comes from book data—Amazon, Bowker, Overcat, library records and CK. It's a very complex weighing of factors—the data always disagrees—which then "bubbles up" from ISBN to work. In its broadest outline, it's the date the most book records say it is. But it gets complex from there." ( )

2023 (when the author page was updated): "Note: Displaying the dates for works is a new feature. Some of the data is not accurate, but updating the original publication date in the Common Knowledge data for the work will update the dates displayed on these pages. The data is updated weekly. Thank you for your help and patience." ( )

Editado: Dez 20, 2023, 12:27 pm

Okay. But there's something strange is going on in some of these cases that seems worth fixing generally instead of just updating CK. I'm not sure I can check Bowker, but Worldcat, Amazon, Overcat, and Google Books say only 1980 for the two ISBNs listed in the LT editions for my example above.

I think this is a good example to investigate because of the small number of copies and editions.

(Note on my example—The actual original publication date should be 1977 for the German edition, Angestellte zwischen Faschismus und Demokratie. But the only German copy in LT is not currently combined with the English copies, so I doubt LT knows that.)

Editado: Dez 20, 2023, 11:57 am

>6 norabelle414:
They adjusted the recalculation time when a date is added to CK, as part of bug fixing on the new author page, to become instantaneous. That was done here:

Jan 8, 9:37 pm

Someone put in the CK date on my example above, sigh. The automatic dates on the Ronald Lewin books from the original post are still incorrect.