ps: the distribution of the full text availability for the sample
considered was as follows:

No preview: 797 (93.5%)
Partial preview: 53 (6.2%)
Full text: 2 (0.2%)

 - Godmar

On Tue, May 6, 2008 at 6:09 PM, Godmar Back <[log in to unmask]> wrote:
> Hi,
>
>  to examine the usability of Google's book viewability API when lookup
>  is done via ISBN, we did some experiments, the results of which I'd
>  like to share. [1]
>
>  For 1000 randomly drawn ISBN from 3,192,809 ISBN extracted from a
>  snapshot of LoC's records [2], Google Books returned results for 852
>  ISBN.  We then downloaded the page that was referred to in the
>  "info_url" parameter of the response (which is the "About" page Google
>  provides) for each result.
>
>  To examine whether Google retrieved the correct book, we checked if
>  the Info page contained the ISBN for which we'd searched. 815 out of
>  852 contained the same ISBN. 37 results referred to a different ISBN
>  than the one searched for.  We examined the 37 results manually: 33
>  referred to a different edition of the book whose ISBN was used to
>  search, as judged by comparing author/title information with OCLC's
>  xISBN service. (We compared the author/title returned by xISBN with
>  the author/title listed on Google's book information page.)  4 records
>  appeared to be misindexed.
>
>  I found the results (85.2% recall and >99% precision, if you allow for
>  the ISBN substitution; with a 3.1% margin of error) surprisingly high.
>
>   - Godmar
>
>  [1] http://top.cs.vt.edu/~gback/gbs-accuracy-study/
>  [2] http://www.archive.org/details/marc_records_scriblio_net
>