I think it would be interesting if Google (or NGCs) developed a way to
prompt the user to disambiguate - Did you mean Mona Lisa (painting) or Mona
Lisa (song)? Preference given to location=Chicago. So that assumptions could
be made explicit, rather than not revealed to the user. But I supose they
wouldn't want to reveal their algorithms that much.
Cindy Harper, Systems Librarian
Colgate University Libraries
charper_at_colgate.edu
315-228-7363
On Mon, Aug 29, 2011 at 10:07 AM, James Weinheimer <
weinheimer.jim.l_at_gmail.com> wrote:
> On 29/08/2011 14:54, Eric Lease Morgan wrote:
> <snip>
>
>> The comment above is very interesting because it begs the question, "To
>> what degree are search results against a database or index expected to be
>> objective?" Our profession has taught us there are right ways and wrong ways
>> to do searching. As a corollary, there must be correct search results and
>> incorrect search results. "If you search the database in the right way, then
>> you will get the correct -- most accurate (precision) and complete (recall)
>> -- results.
>>
>> Yet our profession does not emphasis the inherent characteristics of the
>> reader. (Increasingly I don't like writing the word "user".) If I put in the
>> word "pizza" into Google, I get pizza things close to my geographic
>> location. Our profession does make such assumptions, and we expect the
>> searcher to qualify the query with a location.
>>
>> In this way, Google is easier to use and why different research results
>> will be returned for different searchers. Google sort of knows about you.
>> Ironically, a good librarian will also know about their patrons, and they
>> will create search results tailored for the individual. This is is what
>> reference librarianship is all about. Unfortunately we have yet to migrate
>> this expertise into a computerized environment. "That is artificial
>> intelligence and it can't be done. That threatens me; I will lose my job if
>> that comes to fruition. In order to provide that sort of functionality we
>> will have to record characteristics of readers, and that violates privacy."
>> In short, our own professional ethics have limited us, and others, who don't
>> have these beliefs, have literally profited and grown without them.
>>
> </snip>
>
> These are some of the considerations why I think we need to look at the
> controls in the library catalog as aimed more at librarians (read
> "information expert") instead of members of the general public. I also don't
> think we should label the results of a catalog as more or less "correct" or
> "better" than a result from a full text search engine; that is a treacherous
> path since anybody can justifiably take issue with it. What is "correct"
> and/or "better" is always subjective.
>
> Results from library catalogs should be framed more in terms of
> "standardized" or perhaps "guaranteed". *If* a database has personal name
> authority control, this means that within certain parameters (i.e.
> currently, rule of three, and perhaps practices of analysis) it is possible
> to guarantee that all items authored by certain people can be retrieved by
> the catalog. This is achieved through standardization. The library catalog
> allows standardized methods of finding resources by other concepts, too:
> corporate names, various kinds of titles, subjects, and so on. Full-text
> retrieval tools do not allow this.
>
> The existence of such a tool does not mean that an untrained person can
> retrieve those records successfully, just as an untrained person cannot
> necessarily use a band saw very effectively. They very well may need
> assistance.
>
> Google does not allow any kind of "guaranteed" or "standardized"
> access--just the opposite. If the results vary for you and me, and even vary
> for ourselves depending on where we are searching from, plus it is tweaked
> almost twice a day, I think the public could possibly understand the
> argument for a more standardized means of access.
>
> But I wouldn't call our results better.
>
>
> --
> James Weinheimer weinheimer.jim.l_at_gmail.com
> First Thus: http://catalogingmatters.**blogspot.com/<http://catalogingmatters.blogspot.com/>
> Cooperative Cataloging Rules: http://sites.google.com/site/**
> opencatalogingrules/ <http://sites.google.com/site/opencatalogingrules/>
>
Received on Mon Aug 29 2011 - 10:30:39 EDT