I think you've cut to the chase on this one and seen the potential.  I
went to one of the roll out presentations at Midwinter on Summon, and
was quite impressed.  As someone who *was* an aggregator of metadata in
the recent past (NSDL in the early part of this decade), I can attest to
the fact that making disparate metadata play well together is not an
easy task.  The benefit of doing it the way Summon does it is threefold:

1. They are dealing directly with each provider, and saving the
libraries from having to do this (this is not a small thing, trust me)
2. They recognize the need and the potential for normalizing and
improving the metadata in aggregate (which is generally cheaper, and
vastly improves the behavior and possibilities for the data)
3. Because they also have data on what journals any particular library
customer has subscribed to, they can customize the product for each
library, ensuring that the library's users aren't served a bunch of
results that they ultimately can't access.

This very much the kind of thing that my colleagues at NSDL were trying
to do, but for a lot of reasons (largely political, unfortunately) were
never able to realize.

There will be lots of challenges for them as they move forward with
this, and I for one will be watching closely.  I did speak to Peter
McCracken after the Midwinter presentation, and pointed out that they
might find dealing with personal names a challenge they might want to
take up sooner, rather than later--remember they have both article
metadata and library metadata being presented together, with vastly
different conventions for names ...


Jonathan Rochkind wrote:
> Dr R. Sanderson wrote:
>> Eric,
>> How is this 'new type' of index any different from an index of 
>> OAI-PMH harvested material?  Which in turn is no different from any 
>> other local search, just a different method of ingesting the data?
>> Sounds like good PR to me, rather than a revolution ;)
> I don't know about "revolution". But your right that this _approach_ 
> is similar to an OAI-PMH approach sure. But getting this approach to 
> actually work reliably and succesfully with published scholarly 
> articles -- is NOT easy.  Most of these publishers and aggregators do 
> not have OAI-PMH feeds. (And even if they did, the typical OAI-PMH 
> feed supplying only OAI-DC data does NOT provide sufficient structured 
> metadata for a search of scholarly content).
> Most of them do not share their metadata freely.  You need to have an 
> individual relationship with each one, and you need workflow in place 
> on your end to make sure you have updated metadata for each one, and 
> you need to deal with the inevitable bugs and bad data from the 
> publishers and vendors, and you need to do some normalizing so that 
> they can all actually live together in an index that works for the user.
> This is not an easy thing to do. I believe that some library 
> consortiums have long histories of trying to do this (OhioLink?); some 
> have given up and no longer try to do it (CDL I think?), because it's 
> very expensive and difficult to do right.
> If SerialSolutions has succeeded in doing it so it works well, and can 
> provide it an affordable cost -- this will, in my opinion be huge.
> It's not the basic technology of having a local index that's 
> "revolutionary".  It's, potentially, applying that technology to this 
> particular case, and actually having it produce something that works 
> well for end-users, and being able to do it at an affordable cost.
> But certainly you don't have to be SerialSolutions to do this. I hope 
> that the product succeeds, and I hope that it gets 'competitors' 
> (library consortial and other vendors) so we have options.
> But if it was so easy to do affordably, we'd have it already, right?
> Jonathan
>> Rob
>> On Tue, 21 Apr 2009, Eric Lease Morgan wrote:
>>> On Apr 21, 2009, at 10:40 AM, Mike Taylor wrote:
>>>> I, and most of the people I've worked with, have been using the terms
>>>> "metasearch", "federated search", "broadcast search" and "distributed
>>>> search" synonymously for years.  Have they now settled down into
>>> But I believe we are also seeing a new type of index manifesting
>>> itself, and this new index has yet to be named. Specifically, I'm
>>> thinking of the index where various types of content is aggregated
>>> into a single index and then queried. For example, instead of
>>> providing a federated search against one or more library catalogs, a
>>> Z39.50 accessible journal article index, a local cache of harvested
>>> OAI content, etc., I think we are beginning to see all of these
>>> content silos (and others) brought together into a single (Solr/
>>> Lucene) index and searched simultaneously. I'm not sure, but I think
>>> this is how Summon works.