[Ads-l] GB false positives [was: Semper Gumby (UNCLASSIFIED)]
ADSGarson O'Toole
adsgarsonotoole at GMAIL.COM
Tue Apr 19 07:24:23 UTC 2016
One mechanism for the creation of false positives in the Google Books
database was discussed on this list previously. If Google discovers
that some review of a book X contains a phrase P then sometimes when a
search is performed for the phrase P the Google Books search engine
will display book X as a match. Key point: Phrase P may be absent in
the book, and yet the search engine will still display X when a search
is performed for P.
Some group of Google computer scientists must have thought that this
was a very clever extension to the search engine. A book was being
shown because of an indirect "higher level" association. There are
situations in which this would be useful.
If you change "Sorted by relevance" to "Sorted by date" I find that
many of the spurious matches are no longer presented.
Spurious matches have been displayed by GB for years. I noticed it
when searching for fake Mark Twain quotations because GB kept claiming
that the fake quotations were present in various books by Mark Twain.
But a direct search in the Twain books always showed that the
quotations were absent.
Garson
On Mon, Apr 18, 2016 at 10:05 PM, Joel Berson <berson at att.net> wrote:
> What I've been noticing is that when a GB search returns very few hits, a number of those at the end of the list have not the slightest presence of the search expression, nor even the remotest chance of having one. (E.g., titles seem completely unrelated to the subject.) I imagine this is also happening when many hits are returned, but no-one has persevered to the ends of such lists.
>
> Whether this is arising from GB's goodheartedness and generosity I have no idea.
>
>
> Joel
>
> From: "Mullins, Bill CIV (US)" <william.d.mullins18.civ at MAIL.MIL>
> To: ADS-L at LISTSERV.UGA.EDU
> Sent: Monday, April 18, 2016 4:28 PM
> Subject: Re: [ADS-L] Semper Gumby (UNCLASSIFIED)
>
> ...
>
> I suspect this is one of those cases where GB is offering a book that they think is close to what you want, rather than actually fitting the search.
>
>
>
> ------------------------------------------------------------
> The American Dialect Society - http://www.americandialect.org
>
------------------------------------------------------------
The American Dialect Society - http://www.americandialect.org
More information about the Ads-l
mailing list