[Web4lib] Re: Google News Timeline

Brian Gray mindspiral at gmail.com
Sun Apr 26 12:54:20 EDT 2009


How do we know that "database seems to have gotten it right by accident"? It
is very easy to believe that a database provider sets "relevance" as page
number order if a user drills down by journal title and specific issue. I
cannot imagine that the better database producers all have a single
relevance ranking for all search strategies, since the search process is
very important to the relevance. It makes perfect sense that the relevance
rank was reset to page number order when people drill down through specific
journal issues.


Brian Gray
mindspiral at gmail.com


On Sun, Apr 26, 2009 at 12:10 PM, <IsisInform at aol.com> wrote:

>
>
> Thanks Thomas, for clarifying my  example.
> Peter, I am not characterizing all  relevance rankings as silly, just this
> particular instance.  But, in terms of the database, this is a  pretty big
> instance.
> I did not use the database’s search  screen.  I went to its list of
> publications and selected a journal title.  In successive drilldowns, the
> screen
> displayed year and then issues  within a year.  When a single  issue’s
> articles were displayed, the database allowed sorting by date, source,  or
> relevance.  None of these choices  are pertinent since all returned
> articles are
> published on the same date, in the  same source, and with equal relevance.
> As it happens, the returned articles  are listed in page number order, as
> they would be in a table of contents.  This is the correct default
> structure
> because someone using this technique presumably wants to look at a table of
>  contents.  However the database  seems to have gotten it right by
> accident.  The results are not presented as being sorted by page number,
> but rather
> by date, source, or relevance.  Sorting by page number is not given as an
> option.  Customers who want to know the actual  arrangement have to figure
> it
> out on their own.
> Other prominent databases offer this  type of drilldown by journal title
> and issue.  The one that does it the best allows  sorting by article title
> or
> page number, which obviously are more valuable  presentation strategies for
> this technique.
> The problem seems to be that the  first database offers the drilldown
> method, but then translates the results into  a visible search string.
>  Therefore
>  its sorting options are those used by a keyword search where results are
> usually  less homogenous.  The other database  acknowledges that looking at
> the contents of one issue is a research technique  requiring applicable
> presentation strategies.
> Thanks,
> Katherine
> ***************************
> Katherine  Bertolucci
> Isis Information Services
> Phoenix, AZ 85001


More information about the Web4lib mailing list