new Google Books glitch

Shapiro, Fred fred.shapiro at YALE.EDU
Wed Mar 28 10:40:46 UTC 2012


Actually, when I go into Advanced Book Search and run searches that I have run many times in the past, it looks like it is behaving much the same as it has.

Fred Shapiro



________________________________________
From: American Dialect Society [ADS-L at LISTSERV.UGA.EDU] on behalf of Shapiro, Fred [fred.shapiro at YALE.EDU]
Sent: Wednesday, March 28, 2012 6:27 AM
To: ADS-L at LISTSERV.UGA.EDU
Subject: Re: new Google Books glitch

This is all pretty distressing for those of us who try to use Google Books to find early uses of words, phrases, quotations, and proverbs.  Is there any kind of "advanced search" or "classic Google Books search" that evades the new fuzziness?

It is beginning to seem that, once Hathi Trust gets past its own search problems and develops a more robust search capability, Hathi Trust will supplant Google Books as a tool for historical-lexicographical research.

Fred Shapiro



________________________________________
From: American Dialect Society [ADS-L at LISTSERV.UGA.EDU] on behalf of Stephen Goranson [goranson at DUKE.EDU]
Sent: Wednesday, March 28, 2012 5:09 AM
To: ADS-L at LISTSERV.UGA.EDU
Subject: Re: new Google Books glitch

Yes, Google Book search has changed.
First, apparently, it returns hits with synonyms of the search words.
Second, apparently, it returns, e.g., phrases found in recent reviews of classic books, listing the book rather than the review as the source.
These attempts to be helpful, to me, are not.

Stephen Goranson
http://www.duke.edu/~goranson
_______________________________________
From: American Dialect Society [ADS-L at LISTSERV.UGA.EDU] on behalf of Garson O'Toole [adsgarsonotoole at GMAIL.COM]
Sent: Tuesday, March 27, 2012 10:04 PM
To: ADS-L at LISTSERV.UGA.EDU
Subject: Re: [ADS-L] new Google Books glitch

The results generated by the Google search engine have, I think,
changed noticeably during the past two months. The match algorithm has
been updated so that it now performs some kind of indirect association
mapping to generate more matches.

When one performs a search looking for a precise string in quotes the
results now include many books that do not contain the quoted string.
This behavior has been mentioned in the past on the ADS list, but now
many more spurious matches are being displayed.

For example, when I search for the string "inflamed with wild notions"
the following items are listed as matches:

The Republic of Plato: Volume 1
Socrates: a translation of the Apology, Crito, and parts of the ...
Clouds Aristophanes, Milton Wylie Humphreys

The string does not appear in these books. But the string is part of a
popular quote that began to appear in the 1960s (I think). It is often
misattributed to Plato:

[Begin excerpt]
What is happening to our young people? They disrespect their elders,
they disobey their parents. They ignore the law. They riot in the
streets inflamed with wild notions. Their morals are decaying. What is
to become of them?
[End excerpt]

So the Google match algorithm now apparently performs some type of
indirect matching. The algorithm may look at the set of matches in the
full Google database and creates some kind of signature. It then
matches items to the signature. That is a wild guess. Whatever
technique is being used it is clever. However, it makes my task more
difficult.

For example, when I search for quotes that are incorrectly credited to
Mark Twain Google now presents matches for several works of Twain.
This occurs despite the fact that the target string is absent in
Twain's oeuvre.

Perhaps other list members have observed this behavior.

Garson

------------------------------------------------------------
The American Dialect Society - http://www.americandialect.org

------------------------------------------------------------
The American Dialect Society - http://www.americandialect.org

------------------------------------------------------------
The American Dialect Society - http://www.americandialect.org

------------------------------------------------------------
The American Dialect Society - http://www.americandialect.org



More information about the Ads-l mailing list