As a kind of follow-up to some discussions on suitable software
solutions for search engines for EAD finding aids I would like to give
you what I myself at least regards as a success story.
First an executive summary of the background story:
Ediffah is a project in which we try to establish a network of
Manuscript and special collections departments at larger Swedish
research libraries. having archival collections. The consortium
consists of The Royal Library (Stockholm), and the libraries at
universities in Gothenburg, Lund and Uppsala. These libraries,
together with the Swedish National Archives, now have fully
operational service up and running on
I'm afraid that everything is in Swedish.
During the project we have developed this service, and also
experimented with different authoring tools. I think all libraries
have opted for NoteTab pro. Most of the records in Ediffah are
automatically translated from a much older database that were once
distributed on CD ROM and are very brief; a did and a controlaccess
basically. We do have a few detailed and quite nice once as well
authored during the project.
Here comes a description of the software solution:
First the search engine, after that I give you hints as to how to
build the web interface.
The Zebra Information Server is a product of the Danish company
Indexdata. Zebra server distributed under GPL. There are binary
distributions for Linux (like RPMs for Red Hat -- Suse Linux has it on
the distribution CD roms). I'm not connected to Indexdata in any way
other than that I've used their products for years, and like most of
them. You can find zebra here
The Zebra server is a fully fledged Z39.50 server and can deliver
things like Marc 21. It is also a fully fledged text retrieval engine
and can do a lot of things like ranked search, sorting and a lot of
other things. You get truncation, word lists combined with boolean AND or
boolean OR without any query parsing followed by boring SQL handicraft.
Phrase searching is there.
In addition to this, it can parse XML files (the indexer is linked to
the expat library). Through a clever mechanism you can therefore
connect bib-1 search attributes (socalled use attributes) to XPath
expressions. You are according to the Z39.50 standard allowed to
define your own attributes above attribute 5000.
My configuration file connecting EAD syntax to search attributes is on.
If you know the XPath languge, I think you can figure out what this
file does. Basically it connects all those xpaths in the left column
with "search fields" in the right column. There is another file
connecting those "search fields" to bib-1 search attributes.
The interesting ones are towards the end. This field has three
columns, the first one is the attribute number, the second contains
their names (the same as in the right column in the abs file mentioned
above). The third contains an aggregation of search fields, allowing
you to saying things that when searching bib-1 attribute with this
number, you should also search in the ones listed. E.g., I have
att 5160 origination 5160,5170,5180,5190,5200
att 5170 origination-corpname
att 5180 origination-famname
att 5190 origination-orgname
att 5200 origination-persname
This means that creator names of the four different categories (with
5170 to 5200) are combined into one (5160).
But I also have
att 1 Personal-name 1,1004,1009,5120,5200,5205
att 2 Corporate-name 2,1005,5170
I'm free to combine search fields in many different ways, and through
these to files I can control how a given EAD syntactical construct
should be searched.
The web gateway:
The web interface can be had very easily if you have modern OPAC having a
Z39.50 gateway with support for XML record syntax. It will require a few
xslt scripts. We have searched databases like this using our VTLS Virtua.
I don't belong the OPAC Team so I cannot give any info on how to do this,
but it is doable.
Our database is on
The only thing it is capable of deliver is EAD XML.
I wrote the gateway used in Ediffah in php. But you can find APIs for
this for java, perl and php.
Based on the Virginia OAI server I've made an Z to OAI gateway, so you can
harvest our service:
This is a bit long, but I hope that this might help someone. I have a
little more info (in English, actually) on
Sigfrid Lundberg, Ph.D., Docent phone: +46 (0)46 222 36 83
Lund University Libraries, Head Office email: [log in to unmask];
PO Box 134, SE-221 00 Lund,Sweden http://sigge.lub.lu.se/