Print

Print


> I am beginning to think about how we at Mass. Historical will make our EAD
> files searchable.  I've examined the search interfaces of many
institutions,
> and I've identified three primary approaches (albeit with many
variations).
> First, some institutions simply offer keyword/boolean searching.  Second,
> some institutions allow researchers to search for names, places, titles,
> subjects, dates, etc, in addition to keyword searching.  Finally, some
> enable searching in sections of finding aids, such as in the scope and
> content note or in the biographical/historical note.
>
> I am curious what goals/factors/ideas/assumptions I should keep in mind as
I
> begin to move forward developing a search interface.  I can see merits to
> all three of the above approaches.  What influenced your decisions when
you
> were setting up your search interface?  Does anyone have a keen sense of
> what search options researchers find must useful and/or comprehend most
> clearly?

I am very interested in feedback on this as well.  I am currently working on
an EAD search tool in Perl using CGI.  What features do users want?  My
approach was to allow the user to pick from some preset 'fields'
(Scope/Content, Geogname, Persname, Date, Abstract, Full-Text, etc.) which
corresponded to various elements or sets of elements.  Is this granularity
useful or would a keyword search do just as well?  I also added the ability
to specify individual years or ranges of years to limit the search (using
the collection-level <unitdate/>)--this slows the program down a bit, and
was a pain to write; is it worth having?

One things I haven't implemented is relevance sorting.  Is that necessary?
If so, what factors would influence each matching finding aid's relevance
score?

My search results display each EAD document's title, unitid, origination,
extent, and abstract.  No effort is made to necessarily display the portion
of the document where the search terms were found (as, say, Google does).
Is this a bad thing?  My thoughts are that the abstract is more revealing
than a mid-finding-aid excerpt, but I'm not entirely certain about this.

(On a semi-related note, if anyone has experience with the Perl XML::XPath
module, please e-mail me.)

Thanks,

--Eli Naeher
   Lower Cape Fear Historical Society