Nice work, Ralph.  That's really slick.  I have all the subject terms in the
solr index, but I would like to eventually integrate the Worldcat data to
make the results more relevant (rather than just sorted alphabetically since
each subject occurs only once).

I have yet to adapt my Orbeon forms to handle dynamic querying of the terms
in Solr, but would like to have that done in the next week or two (I hope!).

Here's a sample of the data:

http://beta.scholarslab.org:9080/solr-1.4/terms?terms.fl=subject&terms.limit=25&terms.prefix=Egy

The terms.prefix parameter displays terms that start with the letters
inputted.  It's extremely fast, so I have no doubt I can use it for
autosuggest per keystroke.

Ethan

On Thu, Dec 10, 2009 at 4:07 PM, LeVan,Ralph <[log in to unmask]> wrote:

> In a rare demonstration of doing as promised, I have a FAST
> AutoSuggester running based on that data.  An HTML demonstration of the
> service can be found at http://orlabs.oclc.org/FAST/autosuggest.html and
> the underlying AutoSuggester is running at
> http://orlabs.oclc.org/FAST/AutoSuggest.  Searches from the HTML go to
> WorldCat.
>
> An example of a suggestion request would be
> http://orlabs.oclc.org/FAST/AutoSuggest?query=0
>
> It returns JSON, which the above HTML link demonstrates the use of.
>
> My AutoSuggester is just a thin interface to an SRU database.  The query
> that comes to the AutoSuggester is turned into an SRU query and the
> resulting record is dropped into the AutoSuggester response.  The link
> to the SRU database for the above query for "0" would be
> http://orlabs.oclc.org/identities/search/AutoSuggestFAST?query=fragment+
> exact+%220%22<http://orlabs.oclc.org/identities/search/AutoSuggestFAST?query=fragment+%0Aexact+%220%22>
>
> I've put up a file containing all the records I loaded into my database.
> The records contain 2 tab delimited fields.  The first field is the key
> to the record and corresponds to the user's keystrokes (e.g. '0' above).
> The second field contains the JSON response which is an ordered array of
> terms.  I load that into the moral equivalent of Lucene and you see the
> results in that SRU search.  The file can be found at
> http://orlabs.oclc.org/FAST/AutoSuggestFAST.zip
>
> Let me know if you have any problems or questions.
>
> Ralph
>
>
> > -----Original Message-----
> > From: LeVan,Ralph
> > Sent: Thursday, December 10, 2009 11:12 AM
> > To: [log in to unmask]
> > Subject: RE: [CODE4LIB] Auto-suggest and the id.loc.gov LCSH web
> service
> >
> > We've made some progress on this topic.
> >
> > I have available a list of our FAST subject headings.  They are
> derived from
> > LCSH and may be of some use.  The folks that produced this file are
> working on
> > producing a similar file for LCSH.
> >
> > The file can be found at http://orlabs.oclc.org/FAST/fastOutput.zip.
> >
> > The file contains tab delimited records.  The first column is the ID
> number of the
> > FAST record that the term comes from.  The second column is the MARC
> > Authorities field that the term came from.  The third column is the
> term itself.  The
> > 4th column is the count of records in WorldCat that are retrieved when
> that term
> > is used in the Keyword index.  The 5th column is the count of records
> in
> > WorldCat that are retrieved when that term is used in the Subject
> index.
> >
> > I expect to have an AutoSuggester built on this today.
> >
> > Ralph
>