From one of the Federated Search vendor's perspective... 

It seems in the broader web world we in the library world have lost "metasearch". That has become the province of those systems (mamma, dogpile, etc.) which search the big web search engines (G,Y,M, etc.) primarily for shoppers and travelers (kayak, mobissimo, etc.) and so on. One of the original differences between these engines and the library/information world ones was that they presented results by Source - not combined. This is still evident in a fashion in the travel sites where you can start multiple search sessions on the individual sites.

We use "Federated Search" for what we do in the library/information space. It equates directly to Jonathan's Broadcast Search which was the original term I used when talking about it about 10 years ago. Broadcast is more descriptive, and I prefer it, but it seems an uphill struggle to get it accepted.

Fed Search has the problem of Ray's definition of Federated, to mean "a bunch of things brought together". It can be broadcast search (real time searching of remote Sources and aggregation of a virtual result set), or searching of a local (to the searcher) index which is composed of material federated from multiple Sources at some previous time. We tend to use the term "Aggregate Index" for this (and for the Summon-type index) Mixed content is almost a given, so that is not an issue. And Federated Search systems have to undertake in real time the normalization and other tasks that Summon will be (presumably) putting into its aggregate index.

A problem in terminology we come across is the use of "local" (notice my careful caveat in its use above). It is used to mean local to the searcher (as in the aggregate/meta index above), or it is used to mean local to the original documents (i.e. at the native Source).

I can't imagine this has done more than confirm that there is no agreed terminology - which we sort of all knew. So we just do a lot of explaining - with pictures - to people.

Peter Noerr


Dr Peter Noerr
CTO, MuseGlobal, Inc.

+1 415 896 6873 (office)
+1 415 793 6547 (mobile)
www.museglobal.com




> -----Original Message-----
> From: Code for Libraries [mailto:[log in to unmask]] On Behalf Of
> Jonathan Rochkind
> Sent: Tuesday, April 21, 2009 08:59
> To: [log in to unmask]
> Subject: Re: [CODE4LIB] Serials Solutions Summon
> 
> Ray Denenberg, Library of Congress wrote:
> >
> > Leaving aside metasearch and broadcast search (terms invented more
> recently)
> > it  is a shame if "federated" has really lost its distinction
> > from"distributed".  Historically, a federated database is one that
> > integrates multiple (autonomous) databases so it is in effect a
> virtual
> > distributed database, though a single database.    I don't think
> that's a
> > hard concept and I don't think it is a trivial distinction.
> >
> 
> For at least 10 years vendors in the library market have been selling
> us
> products called "federated search" which are in fact
> distributed/broadcast search products.
> 
> If you want to reclaim the term "federated" to mean a local index, I
> think you have a losing battle in front of you.
> 
> So I'm sticking with "broadcast search" and "local index".  Sometimes
> you need to use terms invented more recently when the older terms have
> been used ambiguously or contradictorily.  To me, understanding the two
> different techniques and their differences is more important than the
> terminology -- it's just important that the terminology be understood.