 |
 |
I think that Mike Giarlo and Michael Witt used the FAST AutoSuggest as part of their databib project [1]. But are you talking about bringing the data down for a local index? //Ed [1] http://databib.org/ On Tue, Jan 29, 2013 at 4:45 PM, Bill Dueber <[log in to unmask]> wrote: > Has anyone created a nice little wrapper around FAST? I'd like to test out > including FAST subjects in our catalog, but am hoping someone else went > through the work of building the code to do it :-) I know FAST has a web > interface, but I've got about 10M records and would rather use something > local. > > > On Tue, Jan 29, 2013 at 4:36 PM, Ed Summers <[log in to unmask]> wrote: > >> Hi Kyle, >> >> If you are thinking of doing name or subject authority control you >> might want to check out OCLC's VIAF AutoSuggest service [1] and FAST >> AutoSuggest [2]. There are also autosuggest searches for the name and >> subject authority files, that are lightly documented in their >> OpenSearch document [3]. >> >> In general, I really like this approach, and I think it has a lot of >> potential for newer cataloging interfaces. I'll describe two scenarios >> that I'm familiar with, that have worked quite well (so far). Note, >> these aren't IR per-se, but perhaps they will translate to your >> situation. >> >> As part of the National Digital Newspaper Program LC has a simple app >> so that librarians can create "essays" that describe newspapers in >> detail. Rather than making this part of our public website we created >> an Essay Editor as a standalone django app that provides a web based >> editing environment, for authority the essays. Part of this process is >> linking up the essay with the correct newspaper. Rather than load all >> the newspapers that could be described into the Essay Editor, and keep >> them up to date, we exposed an OpenSearch API in the main Chronicling >> America website (where all the newspaper records are loaded and >> maintained) [4]. It has been working quite well so far. >> >> Another example is the jobs.code4lib.org website that allows people to >> enter jobs announcements. I wanted to make sure that it was possible >> to view jobs by organization [5], or skill [6] -- so some form of >> authority control was needed. I ended up using Freebase Suggest [7] >> that makes it quite easy to build simple forms that present users with >> subsets of Freebase entities, depending on what they type. A nice side >> benefit of using Freebase is that you get descriptive text and images >> for the employers and topics for free. It has been working pretty well >> so far. There is a bit of an annoying conflict between the Freebase >> CSS and Twitter Bootstrap, which might be resolved by updating >> Bootstrap. Also, I've noticed Freebase's service slowing down a bit >> lately, which hopefully won't degrade further. >> >> The big caveat here is that these external services are dependencies. >> If they go down, a significant portion of your app might go down to. >> Minimizing this dependency, or allowing things degrade well is good to >> keep in mind. Also, it's worth remembering identifiers (if they are >> available) for the selected matches, so that they can be used for >> linking your data with the external resource. A simple string might >> change. >> >> I hope this helps. Thanks for the question, I think this is an area >> where we can really improve some of our back-office interfaces and >> applications. >> >> //Ed >> >> [1] >> http://www.oclc.org/developer/documentation/virtual-international-authority-file-viaf/request-types#autosuggest >> [2] http://experimental.worldcat.org/fast/assignfast/ >> [3] http://id.loc.gov/authorities/opensearch/ >> [4] http://chroniclingamerica.loc.gov/about/api/#autosuggest >> [5] >> http://jobs.code4lib.org/employer/university-of-illinois-at-urbana-champaign/ >> [6] http://jobs.code4lib.org/jobs/ruby/ >> [7] http://wiki.freebase.com/wiki/Freebase_Suggest >> >> On Tue, Jan 29, 2013 at 11:59 AM, Kyle Banerjee <[log in to unmask]> >> wrote: >> > How are libraries doing this and how well is it working? >> > >> > Most systems that even claim to have authority control simply allow a >> > controlled keyword list. But this does nothing for the see and see also >> > references that are essential for many use cases (people known by many >> > names, entities that change names, merge or whatever over time, etc). >> > >> > The two most obvious solutions to me are to write an app that provides >> this >> > information interactively as the query is typed (requires access to the >> > search box) or to have a record that serves as a disambiguation page >> (might >> > not be noticed by the user for a variety of reasons). Are there other >> > options, and what do you recommend? >> > >> > Thanks, >> > >> > kyle >> > > > > -- > Bill Dueber > Library Systems Programmer > University of Michigan Library