Since the advent of online databases there have been concerns about the different interfaces and software provided by publishers and suppliers. In recent years, the growth in the number of databases and full-text electronic journal services has made this aspect of electronic resource provision even more challenging, particularly for Higher Education institutions.
Just as for the foreseeable future databases are likely to continue to be delivered through a variety of interfaces, it is equally likely that there will be increasing demands from users for simplified access. If we do not find ways to deliver this, our users will continue to avoid using these kinds of databases, especially if there are easier routes for finding information. The much-quoted findings of the JUSTEIS Project found that ‘…search engines and known sites are the first resort for most academic queries…’ and that there was ‘…a low take up for bibliographic checking throughout all groups’. 
The library community is not unaware of this problem and research has been taking place into how it might be overcome. One of the aims has been to find a way to use a single interface for accessing a range of databases, despite the differences in their proprietary interfaces. The idea has been to create a portal that can query different databases at the same time, bringing back the results and presenting them to the user in one unified format.
For example, the elib AGORA project  explored this idea as part of its aim to develop a prototype Hybrid Library Management System (HLMS). This HLMS portal incorporated Fretwell Downing’s VDX software which enabled cross-searching of a range of databases, primarily by querying their z39.50 servers instead of their proprietary interfaces. This way of searching across different databases is not a new idea. The z39.50 protocol  was approved by NISO, way back in 1988, as a minimum standard for all databases to enable the exchange of information in a consistent format.
Seeing the potential value of this technology, a number of commercial library management system (LMS) companies have recently started to develop cross-searching products. However, they have found that whilst z39.50 searching works well in theory, in practice the protocol has not been adopted consistently by database suppliers. Some suppliers do not make their interfaces z39.50 compliant at all; others simply do not permit access to their z39.50 servers.
To overcome the limitations of z39.50, the LMS companies have moved the process forward by starting to develop cross-searching software that can work with other protocols, not just with z39.50. If these alternative protocols have been published, like PubMed’s Entrez protocol, this process is fairly straightforward. However in many cases these protocols are not published and so programs may have to be written to extract the data anyway. One such technique is called ‘page-scraping’ or ‘screen-scraping’ and it involves ‘discovering a query’s syntax through trial and error’  by analysing the URLs and the HTML pages of the proprietary database. This ability to exploit HTTP, XML and SQL protocols opens up the possibility of making a much wider range of databases available for cross-searching.
Several of these cross-searching products have now been released in the UK. Fretwell Downing’s Z-portal was presented for the first time at Online 2001 whilst Ex Libris has been signing up its first UK customers for MetaLib. Other systems are available, including MAP [Millennium Access Plus] from Innovative and Encompass from Endeavor. Here at the University of East Anglia (UEA) we have recently installed Ex Libris’s MetaLib, chosen because of its integration with SFX, and with our forthcoming library management system, ALEPH.
What is MetaLib?
MetaLib provides a single integrated environment for managing our electronic resources, whether these be abstracting and indexing databases, full-text e-journal services, CD-ROMs, library catalogues, information gateways or local collections. Through this portal we can present our users with a choice of electronic resources ‘at a glance’, in a way that was not previously possible using our OPAC or static web pages. If it is true that most users limit their use of electronic resources to those with which they are familiar, then MetaLib helps to address this problem by highlighting other resources too.
At the heart of MetaLib is its cross-searching functionality. This allows the user to search a number of our databases simultaneously through a single interface. The results from this broadcast search can be de-duplicated and presented side-by-side for comparison. We have 13 cross-searchable resources from a range of suppliers including, for example, our OCLC databases (Psycinfo, Geobase and the MLA) and our Ebscohost ones (Academic Search Elite and Business Source Elite). We can also include our existing Dynix library catalogue in this cross-searching functionality.
MetaLib also includes SFX, Ex Libris’s context-sensitive linkage software , adding an additional range of options that provide direct links to related full-text and other electronic services.
What functionality does MetaLib offer?
MetaLib offers two ways to access our electronic resources. Firstly, all of our electronic resources may be accessed directly by clicking on their links, just as they would work on static web pages or from a link on our OPAC. This provides ‘Link-to’ functionality through what Ex Libris refers to as the ‘Information Gateway’.
The second way of accessing resources applies to those resources that can be cross-searched using the ‘MetaLib Search’ interface. These are referred to as ‘MetaLib Search’ resources and use what Ex Libris calls the ‘Universal Gateway’. The resources that can be searched in this way still have the ‘Link-to’ functionality as well, so that users have the option of returning to the proprietary interfaces of the databases at any time.
These two ways of accessing our electronic resources are shown on the same screen, so that the resources available via the Universal Gateway and those available via the Information Gateway are presented in a unified way.
Users can start to use MetaLib by selecting one of the ‘Categories’ on the left hand side of the opening screen. We have provided the following categories, mirroring what has previously been available on our electronic resources web pages: Arts/Humanities, Law, Libraries, Medicine/Health Sciences, Newspapers, Reference/General, Social Sciences, Sciences. A list of relevant electronic resources is then presented to the user, with the resources that can use the ‘MetaLib Search’ functionality at the top of the page and those which only have the ‘Link-to’ functionality listed below them.
Adjacent to the title of each resource is an information button, shown as an ‘i’ icon. When the user clicks on this they are shown additional information and metadata about the resource they have chosen. The advantage of this is that the descriptions of resources do not clutter up the page. This data is determined by the library and is entered by library staff using the MetaLib Administration module.
The user can now choose whether to link directly to each of the resources or use the MetaLib search interface to cross-search a selection of these resources.
Cross-searching on MetaLib
To use MetaLib cross-search (its broadcast search facility), users click the check boxes adjacent to the resources of their choice. They then enter their search terms in the ‘MetaLib Search’ boxes at the top of the screen. Options are available to search all fields by keyword or specific fields such as title and author. Boolean searching is also supported and a ‘refine search’ option is available after the first set of results has been received.
The results from the MetaLib search are presented initially with one tab for each database, along with a summary showing the number of hits per resource. This allows the user to access summary results from one database even if the searches of the other databases have not yet been completed. Once the searches have been completed, users click on the ‘Merge’ button to see merged and de-duplicated records (up to a maximum of 150 records).
The summary results are all presented in a unified way in the MetaLib interface. A user will click on the chosen record to view the ‘Full Record’. This full record also offers additional functionality including ‘hypertext linking’ and ‘SFX’, the latter being an integrated part of MetaLib. In terms of hypertext linking, the user can click on the author, title or subject fields to move this data back to the MetaLib search interface, enabling further searches on related terms using this new data. This facilitates further resource discovery.
SFX links provide the user with additional options specific to that record, including links to full-text electronic journal articles where available. After clicking on the SFX button, the SFX menu will appear and this will show the services available for the record: for example, linking to full text journal article or abstract. Other extended services offered by our SFX server include searches on COPAC, title searches on the Internet, and subject searching on the Resource Discovery Network gateways. SFX takes the data from the original record via the Open URL standard and provides these extended services without the user having to retype the data. Any electronic resource searchable using the ‘MetaLib Search’ interface can act as an SFX source, even if it is not normally SFX compliant.
At any stage during a cross-search the user will have the option to return to the proprietary interface. For example, the tabulated summary results may indicate that only one of the chosen resources is really going to be relevant. So the user may choose to link directly to that resource to continue their search. This is yet another way in which MetaLib works as a resource discovery tool, pointing users to the most useful resource in comparison with others.
Users may select records from any of the databases they have cross-searched and transfer them to their ‘e-shelf’ for e-mailing, printing or saving at a later stage. Search queries can also be saved and rerun later in the session or in another session. Users may also create their own Category of resources, called ‘My Resource List’ that allows them to choose their own set of most frequently used resources. This is particularly useful for those working across disciplines. They can then use this list for all their future searches instead of, or as well as, the library-defined Categories. MetaLib also has a ‘Locate Resources’ search function which allows users to search for appropriate resources by Subject or Type, for example, or to choose from a list of ‘All Resources’. This is another way in for users to access the resources most relevant to their studies.
One of the main questions we had prior to the installation of MetaLib at UEA was: how much work has to be done to configure resources so that they can be cross-searched? The good news is that much of the initial work in configuring resources has already been done by Ex Libris or other institutions. This configuration information is accessed through the MetaLib ‘Knowledgebase’, which is essentially a central repository of downloadable configurations, linking rules and settings, which is shared by all MetaLib institutions. As one institution adds a resource, the configuration details are then made available to the others. For resources that are not available this way, Ex Libris also assists institutions with configuration of their resources as part of the installation training for MetaLib. However, there is an expectation that institutions should learn how to configure their own resources, especially z39.50 ones, and this is likely to be a time-consuming process:
‘Because the encoding expected by the different z39.50 servers varies from one z39.50-compatible resource to the next, connecting to a large number of such resources requires meticulous configuration and numerous adjustments.’
This is because each database has its own underlying syntax and logical structure. This means that any search query originating from MetaLib has to be adjusted so that it can ‘talk to’ the target resource. Field mapping is particularly important, especially for author fields, where names and initials may need to be adjusted to take into account syntax differences. The results that come back may also have different logical and cataloguing formats and these will have to be converted so they display correctly in the ‘MetaLib search’ interface.
In our experience, getting to grips with these configuration issues requires input from IT staff as well as librarians, and new technical and pseudo-technical skills may have to be acquired to maintain and develop the range of resources we can offer on MetaLib. At UEA we have found the team approach to configuring resources the only way to make rapid progress, with library IT staff, two Assistant Librarians and the Senior Cataloguer all involved.
The other factor to bear in mind is that when adding resources for the first time, permission has to be sought from the database providers concerned along with information on how to access their servers. Again the expectation is that the individual institution will take the lead in this.
In terms of authentication, MetaLib allows multiple levels of access to be set up. This ‘group functionality’ provides authorisation for the use of certain resources depending on the profile of the user.
At UEA we have kept things simple at this stage by setting up just two options: ‘guest access’ and ‘login access’. When library staff add electronic resources to MetaLib, they make distinctions between resources by cataloguing them as either ‘restricted’ or ‘free’ resources. This ensures that the cross-search functionality of restricted resources will only appear to users on-campus or to users who have logged in, i.e. those who are registered staff or students. This is essential to comply with the terms of our license agreements. The advantage of this authentication system is that it does allow our legitimate users to access the full cross-search functionality of MetaLib both on-campus (via IP) and off-campus (via a username and password).
The ‘link to’ functionality is retained for all resources. So those logging in from off-campus as guest users will still be able to see all the resources that are available, including our restricted ones, but they will be challenged for ATHENS or other authentication as necessary. Off-campus guest users will also be able to cross-search any free resources, such as library catalogues, PubMed, etc., but cross-searching options for our licensed databases will not show.
To avoid any need to issue users with yet another set of usernames and passwords, we are hoping to set up MetaLib to authenticate against our university-wide LDAP (Lightweight Directory Access Protocol) authentication system, so users can use their existing campus-wide usernames and passwords when logging in.
The major outstanding problem is how to include electronic resources that can only be accessed via ATHENS and not via IP address recognition. For example, the ISI Web of Science service cannot currently be included in ‘MetaLib Search’. This is not because the database uses an alternative protocol, since we have already noted that such a limitation could be overcome. Rather it is because, with no other server available, the only route to access the service is via its proprietary interface and that requires ATHENS authentication up-front in the UK. Therefore a direct search cannot be done without having to login to the Web of Science using an ATHENS username and password.
So to include Web of Science, we will need to ask ISI to change their policy about not allowing IP-only access and then we will need their permission to exploit their database for cross-searching purposes. It has been recognised for some time that ‘…effective provision of cross-searching services requires dialogue and co-operation between data provider and library service.’ As one of the first institutions in the UK to install MetaLib, we will no doubt be at the forefront of these kinds of negotiations, but it may be that in future this will be done at national level.
However, there are positive signs that other providers are beginning to allow their resources to be accessed for cross-searching purposes. For example, in February 2002, Ovid Biomed expressed a willingness to enable access to their z39.50 servers, possibly from summer 2002. This should allow us to use resources like Medline, CINAHL and EMBASE in ‘MetaLib Search’.
The other major issue is the ongoing maintenance of ‘MetaLib Search’ resources. Any change by the supplier to database structure or, if relevant, the proprietary interface, may affect existing configurations. This may happen without notice and put the resource concerned out of action as far as ‘MetaLib Search’ is concerned. People can of course still link to the resource directly, but the cross-searching functionality is temporarily lost. In our experience Ex Libris will help with trouble-shooting but it may take several days to resolve such problems.
Another limitation is the unreliability of z39.50 servers. Many database suppliers impose limits on the number of users that can access their z39.50 server at any one time or limit the number of records that can be returned. If this happens, the error messages that display in MetaLib are not very helpful to the user. This is because they do not explain why the search has failed or indicate what the user should do to rectify the problem (e.g. narrow their search). We will also need to make users aware that cross-searching is likely to be less accurate than carrying out the same search via the proprietary interfaces. In particular, author field conversions may be far from perfect. However, as previously mentioned, MetaLib does encourage direct linking so that users can take advantage of the full functionality of the original database.
You will notice that the limitations outlined above have less to do with MetaLib itself than with the availability and reliability of the databases that can be used. These challenges will have to be addressed by other cross-searching portals too.
At the time of writing, feedback from Metalib preview sessions has been very positive and we have just opened up MetaLib to the whole university community for further evaluation and feedback. By September, we hope to have fully implemented MetaLib, including integration with our new ALEPH library management system.
In terms of configuration, we are expecting the process to become less time-consuming as more institutions purchase cross-searching software. Whether institutions choose MetaLib or other portals, the demand for compatible databases is likely to increase. The greater this demand, the more likely it is that suppliers will open up their databases to be accessed in this way. Hopefully the maintenance of z39.50 servers will also be improved, and any restrictions in terms of maximum numbers of users and records will be removed. There is no need for suppliers to be concerned that their proprietary interfaces will be bypassed, because there is always the option for the user to return to these at any time. Indeed the resource discovery functionality of MetaLib may well encourage increased usage of the proprietary interfaces as well.
Although it is early days for MetaLib in the UK, it is encouraging to note that interoperability issues are now on the list of criteria that JISC (Joint Information Systems Committee) takes into account in negotiations for new deals. The issues are also important to those charged with overseeing the integration of electronic resources into the DNER (Distributed National Electronic Resource): ‘The aspiration here is to make more services available through structured protocols such as z39.50, LDAP, Open Archives Initiative. In that way, their data is more directly accessible and manipulable for others to use.’ 
Our next step at the University of East Anglia is to monitor our users’ impressions of MetaLib and respond to their comments. In the meantime, we will continue to lobby publishers and suppliers at both individual and national level to open up their resources for cross-searching. We are also looking forward to exploring the options that will come when MetaLib and SFX are fully integrated with ALEPH, our forthcoming library management system. This will open up new possibilities, such as direct links from MetaLib to an inter-lending module.
It must be remembered that the implementation of any hybrid library product is going to be an ongoing commitment, requiring considerable technical and administrative resources. However there is no doubting the ‘added value’ that a product like MetaLib provides, with its potential to transform the ways in which we have previously used and presented our electronic resources. In this respect, we may indeed find ourselves talking about a revolution…
 Armstrong C.J., Lonsdale, R.E., Stoker, D.A. & Urquhart C.J. 2000. JUSTEIS: JISC Usage Surveys: Trends in Electronic Information Services, Final Report - 1999/2000 Cycle. Department of Information and Library Studies, University of Wales Aberystwyth and Centre for Information Quality Management. The Executive Summary is at: http://www.dil.aber.ac.uk/dils/research/justeis/cyc1rep1.HTM
 For further details about the Agora Project see: Palmer D., Robinson B., & Russell R. 2000. Agora - from information maze to market’. Ariadne, 24 http://www.ariadne.ac.uk /issue24/agora/
 The definition of z39.50 from the Library of Congress, the official z39.50 maintenance Agency, is:
‘Z39.50 is a national and international (ISO 23950) standard defining a protocol for computer-to-computer information retrieval. Z39.50 makes it possible for a user in one system to search and retrieve information from other computer systems (that have also implemented Z39.50) without knowing the search syntax that is used by those other systems. Z39.50 was originally approved by the National Information Standards Organization (NISO) in 1988.’ http://www.loc.gov/z3950/
 Sadeh T. MetaLib and SFX: Managing Heterogeneous Resources in the Scholarly Environment. Conference paper given at CASLIN 2001: Library of Academy of Sciences of Czech Republic and National Library of the Czech Republic, Hotel Na Ostrov, Na Ostrov 816, Beroun 3, Czech Republic, 27 - 31 May, 2001 http://www.caslin.cz:7777/caslin01/sbornik/MetaLib.html
 For further details about SFX see Walker, J. 2001. Linking is as easy as SFX. Library Association Record. 103 (12), 744-746, or http://www.sfxit.com
 Sadeh T. op.cit.
 Pinfield S, & Dempsey L. 2001. The Distributed National Electronic Resource (DNER) and the hybrid library. Ariadne 26 http://www.ariadne.ac.uk /issue26/dner/