Automation and electronic information services are not newcomers to the library world. Back in the 1960s the early library automation systems were already beginning to prove their worth and the development of the MARC record format was well underway. The intervening thirty years have seen the power and features provided by library automation systems improve tremendously, the advent of online services, CD-ROM databases, the rise and growth of the Internet and its associated World Wide Web and beginnings of a truly digital virtual library appear.
Recently the UK eLib programme's projects have investigated many aspects of providing services to an "electronic library". These projects, and other similar initiatives elsewhere, have demonstrated the feasibility of providing catalogues of quality assessed network resources, multimedia electronic journals, online document delivery and digitised short loan material.
Some of these projects have been constructed so as to form part of the operational services of a conventional library but many have been testbeds that exist in isolation, with only a hyperlink from a few conventional libraries' "electronic services" webpages. To the user these services are additional routes that they have to investigate when looking for information. This means that it takes the users additional time and effort and sometimes requires that they go out of their way to locate these services. Whilst many of the services have proved popular, surely we shouldn't demand that the users have to locate and then individually learn to use many different services just to locate the information that they want?
What is needed now is a period of integration and consolidation. eLib and the other digital library development projects have taught us alot about what services we might usefully offer electronically. These services now need to be folded back into the operational systems within libraries in such a way as to make them as easy to use as possible for the end users. After all, libraries are really about providing easy access to information, not providing access to other services.
This is where the concept of the hybrid library comes in. The hybrid library is a library which can provide a "one stop shop" for both hardcopy and electronic resources. Its information systems should provide the end user with a seamless interface that will allow them to locate paper books and journals held locally and at neighbouring sites at the same time as being able to find relevant online resources, electronic publications and digitised material. To do this, the user needs to be provided with a front end that can access information in a variety of databases which are widely distributed and can contain a variety of information in different formats.
Like library automation and electronic libraries, the concept of the hybrid library is not a new one. People have talked about providing access to multiple information systems with a single front end for many years. Often library systems vendors would demonstrate technology that allowed users of their OPACs to access a limited range of other databases and services through the same interface. However now we have the network infrastructure, open standards and available information systems to attempt to build a distributed searching and retrieval system that does not lock people into one vendor's system.
A number of existing technologies are likely to be used in providing the hybrid library. Obviously the Internet and the World Wide Web will have to play a large part as they are now becoming the ubiquitous means of electronic information delivery. Many of the existing services that will need to be tied into the hybrid library only exist on the web. The hypermedia nature of the Web allows it to act as both the front end of the hybrid library's search system if necessary and also the means of delivery of any electronic resources that the end user locates.
To provide the integrated searching over multiple databases, many people are looking to Z39.50 to provide a solution. Z39.50 is an American search and retrieval standard that has been for over a decade. It potentially allows a single search query to be sent to multiple databases and have results that are held in many different internal formats presented to the end users in a single, uniform display. Many commercial library systems now include a Z39.50 server as either an integral part of their OPAC or as an optional bolt-on feature. A number of free Z39.50 clients and toolkits also exist and there are Web-to-Z39.50 gateways already in use.
Z39.50 is a very complex standard and has traditionally been geared towards bibliographic searches. To date this has severely limited its impact on searching for electronic information on the Web. However there is an initiative known as ZSTARTS which intends to make a cut down form of Z39.50 suitable for accessing Web search engines available. If this works, then it might be possible for Web searches to be integrated into the hybrid library's search system as well.
Although Z39.50 offers the ability to search multiple databases it does have one very major drawback. It doesn't include any way of routing a user's query to only the database servers that are likely to have relevant information. Whilst this isn't a problem if you've only got a couple of databases, it is more of a worry if your hybrid library is offering a single front end to maybe hundreds of different databases, spread all over the world. If the system blindly sent each user's query to all of the available databases, the network would quickly grind to a halt!
What is needed is some "forward knowledge" that allows the query to be routed forward to databases that are likely to have useful information, whilst ignoring databases that are known not to have anything relevant. A technology that is currently being developed to allow this is the Common Indexing Protocol (CIP). CIP has grown out of the "centroids" mechanism that is part of the WHOIS++ protocol used by a number of the eLib ANR services. CIP allows databases to provide each other with what are effectively summaries of the information that they hold. Databases that swap this information become part of a mesh around which client applications can route users' queries using referrals from one server to another. This means that the databases that are actually queried for a single search will vary depending on what was in the query (and also maybe other factors such as where the user comes from and what options he has selected). It also means that library systems staff won't need to spend alot of time constantly reconfiguring which Z39.50 servers their clients should be able to talk to (systems staff have more than enough to do already!).
Of course there is still a lot of work to be done before the hybrid library can be fully deployed. CIP must be applied to Z39.50 servers and tested to see how well it scales to cross-domain searching. Web browsers really need to have Z39.50 functionality integrated into them so that the end users have a single front end application to cover all of their electronic information search and retrieval needs. CD-ROM and online database vendors need to embrace open systems technology to allow their products to be searched using the standard Z39.50 search mechanism. Cataloguers need to include online resources into the existing library OPACs.
Bits of this are already happening on a piecemeal basis in many libraries around the world. Hopefully the eLib Hybrid Library projects will provide the push to enable the technologies to be integrated and the vision realised. Then the users and librarians can get on with what libraries are really there for accessing information.