Anything that moves us closer to a methodology for making unified discovery across different collections a routine reality is worth celebrating. I wholeheartedly agree with Chuck on that vision, and am delighted to see what MWeb has done, and much looking forward to future developments Jay hints at. These projects and products demonstrate the immense power of bringing data together, and hopefully will motivate the museum community as a whole to take a long hard look. If federated search turns out to be a useful tool to achieving the vision of searching across many museum collections at a time, I'm on board.
At the same time, I think it is also worth having a look at what others who have pursued this path have to say. Here are some of the main issues with federated searching as gleaned from the library experience with it. At first glance, these seem to apply in a museum context as well. * You usually have to wait until all targets respond, so your wait-time for a search result can be fairly long in a federated search. * You usually only receive a specific subset of records with hits from each target at a time. * Since you only have a subset of actual records with hits in hand, you are fairly restricted in your ability to offer faceted browsing. * You usually have to depend on the ranking the original target provides you, and to harmonize rankings between results from different targets is a challenge. To put it simply, it's difficult to assess whether the first hit from database A is as important as the first hit in database B, or whether the first 5 hits from database A are much more relevant than the first hit from database B. That means you are usually restricted to just show hits segregated by the different databases, not integrated into a single run-on search result. In the UK, the national museums and galleries have just launched Creative Spaces (http://bm.nmolp.org/creativespaces/), which also relies on federated search, and there have been discussions about the pros and cons of federated search in that context as well. I'd be happy to learn more about how these issues can be addressed from Stephen and Jay. For those who have the patience to wade into another community's discourse to learn lessons for our own, I'll recommend this article by Jonathan Rochkind, which gives a good background to what I've tried to summarize above: http://www.libraryjournal.com/article/CA6413442.html Cheers, G?nter -----Original Message----- From: mcn-l-bounces at mcn.edu [mailto:[email protected]] On Behalf Of Chuck Patch Sent: Friday, March 13, 2009 12:57 PM To: Museum Computer Network Listserv Subject: Re: [MCN-L] MWeb Universal I agree with Jay. And let's not forget that the market this demonstration piece (PastPerfect users) showcases is one where federated search is something both truly unusual and highly beneficial given the frequency with which these collections are small and not easily accessible. Chuck Patch On Fri, Mar 13, 2009 at 3:41 PM, Jay Hoffman <Jay at gallerysystems.com> wrote: > Dear Stephen, G?nter et al, > > First, congratulations to Stephen for making an interesting project > available. > > I would like to weigh in on this conversation, lest everyone reading it > think that the statements about federated searching should be accepted > without challenge. > > I don't agree that one should assume that "enthusiasm about federated > searching ... has dissipated by now". Maybe "for now" would be a more > appropriate way to put it. There seems to be so many experts with so much > experience in this field in the museum informatics space (this is another > tongue-in-cheek statement). > > There is still research and development taking place in > federated/distributed searching and more interesting things to come, > including ranking. > > Stephen's project is an interesting step in the right direction and I think > we should look at the benefits of the architecture before being dismissive. > The notion that the technology might only be good for a "limited number of > data sources" will also soon be dispelled. > > Best, > > Jay > > Jay Hoffman, CEO > Gallery Systems > 261 West 35th Street, 12th Floor > New York, NY 10001 > jay at gallerysystems.com > +1.646.733.2239 > > > > > > > -----Original Message----- > From: mcn-l-bounces at mcn.edu [mailto:mcn-l-bounces at mcn.edu] On Behalf Of > Waibel,Guenter > Sent: Friday, March 13, 2009 12:41 PM > To: toney at systemsplanning.com > Cc: Museum Computer Network Listserv > Subject: Re: [MCN-L] MWeb Universal > > Dear Stephen, > > While I don't mean to rain on your parade, I have to admit that my question > was a little bit tongue-in-cheek. The library community has been trying to > get meaningful ranking of federated search results to work for the better > part of 10 years now. At this point, it's widely acknowledged that ranking > is the Achilles heel of federated search, and even with the best technology, > a limitation which can't be completely mediated. I'd say that a lot of the > early enthusiasm about federated searching as the solution to integrating > datasources, especially at scale, has dissipated by now. > > Having said that, I can see the usefulness of this kind of search for a > limited number of datasources, such as library, archive, museum collections. > It's a great first step to see all the content in one place! > > Cheers, > > G?nter > > > -----Original Message----- > From: Stephen Toney [mailto:toney at systemsplanning.com] > Sent: Friday, March 13, 2009 8:40 AM > To: Waibel,Guenter > Cc: Museum Computer Network Listserv > Subject: RE: [MCN-L] MWeb Universal > > Hello Gunter, > > At present MWeb Universal sorts alphabetically by the brief name of the > database, subsorted by the type of content. Ranking is a good idea and > we will add it to a future release. > > Thanks for the suggestion! > Stephen > > - > Stephen Toney > Systems Planning > toney at systemsplanning.com > http://systemsplanning.com > > MWeb, CAPS, MARCView, and MARConvert are trademarks of Systems Planning > > > On Fri, 2009-03-13 at 11:29 -0400, Waibel,Guenter wrote: > > Hi Stephen, > > > > How do you deal with ranking search results from various sources? > > > > The Baltimore example seems to side-step ranking. It displays hits > segregated by database. > > > > Cheers, > > > > G?nter > > > > -----Original Message----- > > From: mcn-l-bounces at mcn.edu [mailto:mcn-l-bounces at mcn.edu] On Behalf > > Of > Stephen Toney > > Sent: Friday, March 13, 2009 7:03 AM > > To: Chuck Patch > > Cc: Museum Computer Network Listserv > > Subject: Re: [MCN-L] MWeb Universal > > > > > > On Thu, 2009-03-12 at 09:57 -0400, Chuck Patch wrote: > > > Hi Stephen, > > > > > > This is really interesting. ... looks like you are inches away from > > > being able to create a PastPerfect Consortium. > > > > > > Thanks, Chuck, > > > > Actually MWeb Universal can do that now. Try > > http://searchbaltimore.pastperfect-online.com > > to search a consortium of three Baltimore museums. > > > > The current release is suitable for consortia of any museums, not just > > PastPerfect sites, since it can search any CMSs or databases without > > exporting, FTPing or Z39.50. > > > > MWeb Universal has been tested with 50 databases at once. > > > > Thanks! > > Stephen > > - > > Stephen Toney > > Systems Planning > > toney at systemsplanning.com > > http://systemsplanning.com > > > > MWeb, CAPS, MARCView, and MARConvert are trademarks of Systems Planning > > > > > > > > _______________________________________________ > > You are currently subscribed to mcn-l, the listserv of the Museum > Computer Network (http://www.mcn.edu) > > > > To post to this list, send messages to: mcn-l at mcn.edu > > > > To unsubscribe or change mcn-l delivery options visit: > > http://toronto.mediatrope.com/mailman/listinfo/mcn-l > > > > The MCN-L archives can be found at: > > http://toronto.mediatrope.com/pipermail/mcn-l/ > > > > > > > > _______________________________________________ > You are currently subscribed to mcn-l, the listserv of the Museum Computer > Network (http://www.mcn.edu) > > To post to this list, send messages to: mcn-l at mcn.edu > > To unsubscribe or change mcn-l delivery options visit: > http://toronto.mediatrope.com/mailman/listinfo/mcn-l > > The MCN-L archives can be found at: > http://toronto.mediatrope.com/pipermail/mcn-l/ > > _______________________________________________ > You are currently subscribed to mcn-l, the listserv of the Museum Computer > Network (http://www.mcn.edu) > > To post to this list, send messages to: mcn-l at mcn.edu > > To unsubscribe or change mcn-l delivery options visit: > http://toronto.mediatrope.com/mailman/listinfo/mcn-l > > The MCN-L archives can be found at: > http://toronto.mediatrope.com/pipermail/mcn-l/ > _______________________________________________ You are currently subscribed to mcn-l, the listserv of the Museum Computer Network (http://www.mcn.edu) To post to this list, send messages to: mcn-l at mcn.edu To unsubscribe or change mcn-l delivery options visit: http://toronto.mediatrope.com/mailman/listinfo/mcn-l The MCN-L archives can be found at: http://toronto.mediatrope.com/pipermail/mcn-l/
