Conference papers

Digital collection management

Digital collection management - building a digital science library

Kutira Bandte, Anne Camfield, Philip G. Kent and Rolfe Westwood

CSIRO Information Technology Services, Clayton, Victoria

Abstract

CSIRO, our premier science organisation, is building Australia's largest digital library of full text science journals. The philosophy behind the decision to use the ScienceServer software that supports local loading of journals is discussed. CSIRO's technical architecture including the implementation of Storage Area Network (SAN) technologies to handle multiple terabytes of storage is described. Political and implementation issues such as funding, transition from print to electronic, authentication, training, documentation and support are expanded upon. CSIRO's plans for the future including aggregation of further content, integration with other secondary information sources and the 'Holy Grail' of a single interface are discussed. The potential to provide access to other Australian organisations is also noted.

Introduction

Australia's Commonwealth Scientific and Industrial Research Organisation (CSIRO) is an independent statutory authority constituted and operating under the provisions of the Science and Industry Research Act 1949. As Australia's premier science agency, CSIRO has a vision:

To be a world-class research organisation vital to Australia's future.

The organisation has some 6400 staff working in laboratories and field stations around Australia. Their work covers a broad range of areas of economic or social value to the nation, including agriculture, minerals and energy, manufacturing, communications, construction, health and the environment.

A knowledge-rich organisation, CSIRO's research has always been heavily supported by library and information services, with the majority of literature being serial in nature. While library services are conducted in a devolved manner across CSIRO, corporate leadership and coordination for CSIRO's Library Network are provided through CSIRO IT Services. The Organisation has felt the impact of increasing serials prices and negative exchange rates in recent years through a decreasing number of serials to support the scientific endeavour. Table 1 evidences this impact.

Table 1: CSIRO library network serial acquisitions 1994-8

(Note: While these figures represent centralised serials puchasing in CSIRO, they do not include the devolution of some serials acquisitions to individual CSIRO libraries in 1997)

Following the Priorities and Pathways: Review of CSIRO Library Network in 1997, CSIRO's executive committee endorsed the recommendation that high priority be given to 'coordinating the delivery of quality electronic information to all CSIRO staff'.

Some progress had been made in delivering electronic information resources within CSIRO prior to that date. However, subsequent to this milestone, a concentrated effort was made to achieving the vision of delivering as much information as possible to the scientist's desktop.

Some aspects of the Organisation's ability to move forward with the digital library concept included:

• Organisational willingness to invest in information technology infrastructure
• High level support from CSIRO's Executive
• An integrated approach to corporate Information Technology and Information Management with leadership from Jonathan Potter, General Manager, CSIRO IT Services
• Support within research divisions and the key role of Dr Graham Harris, Chief CSIRO Land and Water and Chair of CSIRO's peak library committee, the Library and Information Access Group (LIAG).

A co-ordinated approach to purchasing

CSIRO has always acquired its scientific serials through a centralised purchasing function, achieving economies of scale and expertise. Corporate contracts sourced these print serials from a small number of subscription agents. The move to electronic serials did not necessarily offer reductions in the purchase price but CSIRO saw an opportunity to leverage its existing expenditure to obtain additional titles and to facilitate access across the whole organisation, including remote sites.

An increasingly multidisciplinary approach to science means that scientists could benefit from CSIRO-wide access and previous duplication of titles could be exchanged for new titles.

Our purchasing records demonstrated that CSIRO's most significant tranche of commercial science journals emanated from Elsevier Science. Approximately 635 subscriptions to 415 unique Elsevier Science titles worth $AUD 1.7 million were acquired in 1999. Elsevier Science was keen to explore new ways of working with CSIRO, one of their major customers in Australia. Despite negative concerns within the library industry about Elsevier's market share and price profile, the Elsevier Science collection offered CSIRO an opportunity to achieve a high impact through the conversion of these titles to electronic delivery.

CSIRO's sister agency the Australian Nuclear Science and Technology Organisation (ANSTO) provides a joint library facility for both ANSTO and CSIRO staff located at the Lucas Heights site. ANSTO has been a member of the CSIRO Library Network for many years, sharing services and infrastructure. In addition, CSIRO also provides library services to Food Science Australia (FSA), an unincorporated joint venture between CSIRO and the Australian Food Industry Science Centre (AFISC). As both ANSTO and FSA print serials were already acquired through CSIRO, the total subscription list became the basis of negotiations.

CSIRO's stance through direct negotiations with publishers and in using a professional negotiator also differs from the general library industry. Commercial science publishing is dominated by a small number of major publishers. Through agreements with this group of publishers the majority of commercial science titles may be acquired. It is acknowledged that this approach may not be as appropriate for small or societal presses or indeed for broader disciplines with a large number of publishers.

The ScienceServer solution

CSIRO has a preference for the purchase of off-the-shelf software solutions. A complete package that provided software for the delivery of electronic content was an attractive proposition.

Elsevier Science has a joint venture interest in ScienceServer LLC with Orion Scientific Systems, a company with a profile in web-based information retrieval and analysis systems for government and industry. ScienceServer had an established customer base in organisations with a similar profile to CSIRO such as Los Alamos National Laboratories, New Mexico,USA, the Ohio Library and Information Network (OhioLINK) of academic libraries in Ohio, USA, and the library of the University of Toronto, Canada. Discussions with representatives of these sites, Orion and ScienceServer and Elsevier Science provided useful input to CSIRO's decision to proceed with the software acquisition and subsequent implementation.

This formed the basis of close links with other Scienceserver sites that have continued forward after implementation. Of particular note is CSIRO's close affinity with Los Alamos National Laboratories (LANL). Rick Luce, Director of LANL's Research Library and Library Without Walls Project is a visionary and has a significant achievement in this area. CSIRO has leveraged the experience of Rick Luce and other ScienceServer sites to assist in our publisher strategy and to reduce reinvention at CSIRO. Further information on the LANL experience is documented by Pack and Pemberton (1999).

Why load locally?

The ScienceServer solution provides local loading of publisher content on the customer's hardware. Why did CSIRO choose this local loading solution instead of going direct to multiple publisher's sites, such as Elsevier's ScienceDirect internet service? Many reasons specific to CSIRO include:

• Archiving - perpetual access to all loaded data even if subscriptions are cancelled.
• Access - transfer of large volumes of data (as journal articles) is cheaper and faster when served from within Australia under current Australian Academic and Research Network (AARNet) pricing arrangements
• Integration - we can choose what we add and integrate into the collection. CSIRO now has the knowledge to develop loaders utilising content from other publishers and secondary data sources.
• Single point of access and common front end
• Security and privacy - CSIRO has access control within our own network. As CSIRO conducts research of national significance and sensitivity, a local solution provides greater security for usage data.
• Sovereignty - an Australian source for the data.

CSIRO IT services also had the network infrastructure, technical expertise, help desk facilities to support such a venture.

There are however disadvantages:

• CSIRO must negotiate with each publisher
• Not a cheap option
• Not all publishers will provide full content for local loading

The electronic journals working group and e-journal trials

As part of CSIRO's priority to deliver electronic information to CSIRO staff, LIAG convenes specialised working groups according to specific needs of the library network. The Electronic Journals Working Group (EJWG) was established as a result of a recommendation from the Library Network Review in November 1997. It comprised staff from CSIRO libraries and ITS, information management.

A major role of the EJWG was to maintain a 'watching brief' on the evolution of electronic journals, ensuring that the CSIRO Library Network was kept informed of developments.

During the life of the working group, ITS obtained and installed an evaluation copy of the ScienceServer software and a selection of Elsevier titles. The working group assisted in the subsequent trial of the software and Elsevier content that ran from 19 October - 18 November 1998.

The working group undertook a coordinating role for the evaluation, managing enquiries, liaison and end-user surveys with several CSIRO libraries. The trial enabled the library Network to address questions and concerns that arose. These were on issues such as timeliness of delivery, quality of electronic issues and the functionality of the ScienceServer software.

Many of the questions raised during the trial were either answered by ITS staff, or passed on to ScienceServer technical support or Elsevier for clarification. The answers became the basis of a Q and A web page made available to the Network.

The trial outcomes were also influential in many software improvements by the developers. This process also helped to develop a strong relationship between the developers and CSIRO that continues to this day.

Integration with the csiro library network catalogue

Prior to the ScienceServer opportunity, a March 1998 audit concluded that CSIRO obtained 435 electronic journals. Librarians across CSIRO's Library Network were cataloguing these in our Integrated Library Management System (ILMS), Voyager. Cataloguing is a devolved activity. The Voyager Online Public Access Catalogue (OPAC) provides automatic links to these journals for Internet Protocol (IP) address authenticated items.

In July 1999, a Voyager Electronic Journals Workshop was held to discuss issues related to electronic journals and specifically to address issues related to the upcoming implementation of the Elsevier full text journals.

A group of twenty cataloguers, reference librarians and corporate staff from across Australia attended the two-day Workshop. Corporate staff included those involved in the negotiations of the ScienceServer system and its content and those administering Voyager.

The workshop's main aim was to clarify cataloguing issues, particularly to decide on the use of bibliographic records. Other aims were to finalise texts for consistent OPAC display, and to gather information about other issues related to electronic resources, (eg how do Divisions inform their scientists of e-resources, are there other more efficient/effective options?).

Issues
The corporate acquisitions group originally raised the issue as to whether there should be separate bibliographic records for an electronic and print subscription. It seemed like an enormous amount of work to create an additional bibliographic record to represent a subscription to the electronic version separate from the subscription to the print journal. The acquisitions staff wanted to use 'proper' bibliographic records when creating and/or modifying subscription orders in the ILMS. A standard CSIRO approach was needed.

There was also much discussion on management of e-journals in general. The workshop participants felt that the corporate group should provide much more guidance towards decisions on purchasing and managing access to e-journals. Basically the corporate role is to manage 'whole of CSIRO' purchases and individual libraries manage their own individual purchases. The corporate group is not resourced to manage all the individual titles.

Solutions

The Workshop decided to concentrate on the cataloguing issues.

The first issue was the number of bibliographic records (ie one for print and one for electronic or one for both). Robert Ho, senior cataloguer, gave an overview of the impact for either option.

The consensus was to use one record for both print and electronic.

These are the guidelines for additions to a Bibliogaphic record for a print version, when adding e-journal information/holdings:

Comprehensive documentation was written (Cataloguing Note CN40 : Electronic Journal Cataloguing). This document is available to CSIRO cataloguers on the web. It includes extensive notes on each of the tags and contains many examples.

One of the main stumbling blocks at the workshop was deciding on the correct place for the URL. Should the URL be in the main bibliographic record or in the holdings record? There was much debate on this issue, particularly when taking into consideration that previous guidelines for the cataloguing of electronic resources was to put this information into Tag 856 in the main bibliographic record. That is, most cataloguing of e-journals in our database already had the URL in the main bibliographic record. If there was a change to this policy, it meant divisional cataloguers would need to amend those records. They were not impressed with the extra workload required of them. To this day, most of this work has not been completed. So the decision was to create both a holdings and bibliographic 856 Tag for Electronic location and access.

This still wasn't accepted unanimously but at least agreement was made. The idea of two URL links displaying in the OPAC was a concern. This was alleviated by the fact that the bibliographic Tag could be masked. Unfortunately, since cataloguers have not completed the extra work required to add 856s to holdings records the URL in the bibliographic record in the OPAC has not been masked. But it is better to have too much access than not enough access.

Holdings record

A new location - CSIROALL was created for holdings records to clearly indicate that access to these holdings was for anyone in the CSIRO Library Network. This location was required as CSIRO did not have a location that represented the whole consortium. Up to now locations were based on a site or collection. This was quite acceptable for print collections where the location represented the physical locality of the item. But in this ethereal world the 'location' was not always directly related to a physical library collection.

Merged holdings for print and electronic holdings

The Workshop agreed that a single merged holding record was a good approach when a location holds both a print and electronic version of a resource. The OPAC could clearly distinguish the print and electronic issues through consistent use of MARC tags and indicators.

The merged record would contain an 852 field for the print location/call number; an 856 field for the link to the URL and any notes about availability of the electronic version; an 866 field for the electronic version holdings; and an 866 for the print version holdings. The two 866 fields have different indicator 2 values and thus allow two different display labels in the OPAC.

On completion of the cataloguing guidelines for e-journals, the Elsevier records were updated in Voyager. This involved about 600 new bibliographic records and 500 current bibliographic records and their holdings statements.

URL checking

There was some discussion of URL-checking; who should do it and how, including could/should it be done centrally. Some librarians use a link checker utility; useful for identifying redirects. An effective and efficient process for managing URL accuracy and resolution is still to be found.

Web based lists

Many CSIRO libraries produce a divisional intranet web page listing electronic resources. To assist libraries in the production of those pages a number of Voyager reports were created based on discussions at the Workshop. These reports are available to librarians directly from the web.

The Voyager Electronic Journals Workshop resulted in a number of timely changes for CSIRO's Library Network. It refined the cataloguing guidelines for electronic versions of journals. It recognised the shift to centralised cataloguing for CSIRO-wide electronic purchases. It raised the level of knowledge of issues involved in the management of electronic resources, at local and corporate levels.

The full suite of Elsevier titles was available through the CSIRO Library Network Catalogue (OPAC) at the launch. CSIRO scientists are able to find all titles in the Electronic Journal Collection (EJC) system through the library system. The EJC may be the CSIRO digital library, but the OPAC continues to be the metadata repository for all CSIRO library collections.

Pre-implementation

In June 1999 CSIRO's Executive Committee approved a decision to proceed with the purchase of the ScienceServer software and to load the Elsevier content. CSIRO's decision was based on successes at other sites worldwide, improvements in the quality of the files that had been a problem in the earlier trial and an ability to negotiate an attractive rate for the complete Elsevier collection. Due to concerns about quality and cultural issues in implementation, the three year agreement was constructed around a transition from print to electronic. In 1999 all print was retained, in 2000 fifty percent of print was retained and in 2001 there is electronic only delivery except where electronic is not available.

As the content was to be extended to other publishers in addition to Elsevier, a generic name: Electronic Journal Collection (EJC) was selected for the service.

With the time and money invested in the project, it was important that the launch of the EJC was successful.

The EJC roll-out preparations took place from June to August 1999. They involved preparing documentation and training materials and developing a marketing and publicity campaign. Presentations were also organised at all CSIRO sites around Australia.

From the very beginning of the project it was clear that CSIRO was introducing 'cultural change', not just an electronic journal system. The EJC would change the way that scientists would access research material and approach their research. Based on the advice of some librarians, IT Services was concerned that many scientists would be anxious about this.

In order to manage the change and successfully deal with cultural change issues several strategies were employed :

Involvement of CSIRO library network

According to standard practice, CSIRO library staff provided invaluable feedback on training materials and documentation before piloting with users. CSIRO librarians were also actively involved in system testing. In addition, they assisted with the logistics for the presentations (eg coordinating local participants, availability of meeting / conference venues, equipment and technical support). The project team held briefing sessions for divisional library staff in most Australian states prior to the end-user presentations. This ensured that the CSIRO librarians were well resourced and informed and felt confident in their role of 'first port of call' for most end users' queries. The presenter was introduced by the local librarian at each session. This demonstrated that the introduction of the EJC was a true 'Library Network Initiative'.

Staff secondments

The team involved in the roll-out essentially consisted of EJC Project staff. In addition several staff were seconded from other projects to assist with the development of training and marketing materials, the planning of the training schedule and to participate in the end-user presentations. Amongst them was a high-profile Divisional Librarian seconded especially to provide a CSIRO library staff and end-user perspective. It was this mix of corporate and divisional expertise that achieved the planned outcome : a sense of good-will amongst EJC users and their preparedness to give the EJC 'a go'. Consequently the composition of the roll-out project team was instrumental in the enthusiastic uptake of the EJC by CSIRO's scientific community.

Marketing and publicity

Our rigorous marketing and publicity strategy was conceived by a project team member who had recently completed MBA studies in marketing and public relations. Through this campaign maximum exposure was created for a system that would change the working life of the majority of CSIRO research staff. Working with a staff member who had a graphic design background, the look and feel of the systems banner and of all documentation, correspondence and promotional material was streamlined. This essentially 'branded' the EJC and ensured that scientists recognised the system by its colour scheme and logo. Two weeks prior to the launch EJC Information Packs were despatched to all Chiefs of Division, Systems Managers, Librarians, the ITS General Manager and Assistant General Managers and other key ITS staff. These information packs contained :

• an introductory letter
• a poster (right)
• a set of quick reference cards (QRCs)
• journals list
• a printed example of an article (this was actually a full-text article matched to the research area of the Division to which the information pack was sent)

Resources and documentation

As part of launch preparations, an 'EJC project homepage' was created on which a wealth of information and resources for Divisional Librarians and CSIRO scientists was available. When first implemented it featured background information on ScienceServer, the EJC Project and general issues surrounding electronic journal use in CSIRO. A 'Documentation' section included a set of 4 Quick Reference Cards (QRCs), a 'Tips and Tricks' series of brief 'cheat sheets', a 'Quick-Start' guide. A 50 page User Guide was also written. However after the final draft it was determined that users did not require this lengthy document. User feedback and helpdesk calls showed that they were happy with the documentation already provided. The EJC Project Homepage has now become the official noticeboard for information about new releases, new features and other news

Launch

The actual launch of the EJC took place on 17 August 1999 and was announced through an e-mail from the CSIRO Chief Executive to all CSIRO staff. Shortly after the launch the road show presentations began.

Road shows

End-user training presentations were held at all major CSIRO sites from 17 August to 10 September 1999. In addition to the EJC poster that had been dispatched with the Information Packs, local librarians were provided with special flyers (see below). These could be pasted over the EJC posters to advertise the time and date etc. of the presentations on each site.

In each of the sessions the presenter provided information about the EJC, addressed cultural change issues, and presented a live demonstration of the system. Although there was some nervousness about live demos (some remote sites don't have considerable bandwidth) it all worked satisfactorily. Presentation participants also received a full set of printed Quick Reference Cards (QRCs).

In addition to end-user documentation, a training kit was developed for librarians wishing to train end-users who were not able to attend any of the presentations or to introduce new staff members to the EJC. These training kits were downloadable from the EJC project homepage and included :

• Pre-Presentation Check List (.doc format)
• Presentation Slide Show (.ppt format)
• Live Demonstration Notes (.doc format)
• Stop Word List (.doc format)

The road show of presentations was followed by a promotional article in the Spring 1999 edition of CSIRO's quarterly staff magazine CoResearch.

The excellent uptake of the EJC and the enthusiasm with which CSIRO scientists embraced the electronic journals proved that the strategy employed for promotion, training and roll-out had been a successful approach. Consequently a similar strategy was used for the roll-out of version 3.1 that took place early in 2000, although Road Shows were not required.

The methods used for the initial EJC roll-out have since become a de-facto standard within IT Services for the introduction of corporate systems into CSIRO.

Technical description

The ScienceServer software has provided many advantages for the end-user, librarian, operations and systems personnel.

User Advantages

For the scientist, the ScienceServer is a web-based service that delivers locally stored journal content in electronic format. The functionality includes the ability to:

• Browse through tables of contents, navigate through journals, issues and articles using hypertext links
• Bookmark ScienceServer URL's to issues and articles for future reference
• Read articles on-line
• Print high-quality versions of articles, including colour
• Search interface enables searching any indexed field such as the title, abstract, author. Furthermore, full text of each article is indexed during the load process and the system allows searching across the full text of all the articles in the collection to find relevant material.
• Save frequently used searches to be re-executed at a later time
• Run automatic searches (SDIs) each time new data is added, and receive the results through e-mail as Table of Contents (ToC) or subject based alerts.

The user interface is intuitive with both browsing and searching easy to use.

Librarians

For the librarians, several other additional features have been developed:

• An e-mail alert system informs of all new issues added in the recent load
• Separate listings detail new titles added to the collection
• Links are available from the issues listing page to holdings and bibliographic details in the library catalogue.
• A web page links to resources such as training material, quick reference cards and other documentation
• A system for claiming/tracking missing issues is available (electronic issues like their print counterpart can fail to be delivered)

Operational/Administration

The system has been designed to keep operational overheads to a minimum and as many of the day-to-day processes as possible have been automated. An example of this is the regular loading of data.

The Elsevier content is delivered on CD, loaded that evening and made available the next day to researchers. The load process is automatic, and the only manual process is the physical insertion of the CD. Various byproducts of the load process such as e-mail messages alerting users of new titles, and listings of titles on the web pages are all automatically generated.

Other features provided by the system include:

• Content loaders to handle the indexing and creation of XML
• A tool kit approach to systems development that has enabled CSIRO to develop its own loaders.
• Fully configurable and customisable system in the form of templates to enable customised header, footers for each web page.
• Entitlements system whereby permission to access a publisher's titles is based on whether consortium members are covered by the Agreement.

Authentication and access

A key decision was to adopt name and password access to the collection rather than Internet Protocol (IP) filtering. While this was more difficult to establish, it has subsequently simplified the profiling of the different groups in the consortium. The original reason for doing this was that name/password access enabled certain features of ScienceServer to be used (eg the ability to save searches; set up e-mail alerts; view search histories). Name and password authentication would also enable the system to be used from anywhere that the user had an internet connection: work; home; while travelling.

This complicated the implementation as CSIRO wanted to avoid the maintenance of yet another database of user names and passwords. There was however a solution.

Many divisions in CSIRO were already using the Microsoft NT4 domain (Nexus) to authenticate their users for access to LAN resources. Approximately half CSIRO personnel had Nexus accounts, and there was an ongoing push for the remainder of the Divisions to convert. The opportunity was there to further this push by using this Nexus domain for authentication.

The Apache server that was chosen as the underlying web server was enhanced by the Unix gurus in ITS to provide this hook into Nexus and provide authentication. That took care of the CSIRO users.

There was a group of non-CSIRO employees who were part of the Library Network but could not (eg security considerations) be added to Nexus. The solution again lay with the Apache web server. An extension enabled authentication against multiple sources and worked as follows.

A request to the server would prompt for name and password. This was checked against Nexus, if this failed, it was checked against an alternate source. In this instance a flat file of 'Unix' style names and encrypted passwords was checked. So, all that was needed was a mechanism to supply names/passwords.

A subsequent system was put into place so that each non-CSIRO group entitled to access would be responsible for maintaining their own name/password lists. ITS specified the format and provided a secure FTP site for such lists. A procedure was set up to automatically process the files. An organisation could update their lists to add, delete or change accounts as often as necessary.

The advantages to CSIRO are:

• Automated process so no manual updating of the external names and passwords
• Access is secure as the name/password only gives the external group access to the EJC and not other parts of CSIRO's intranet
• Name/password maintenance is the responsibility of the external organisation and no extra work would be placed on the IT Helpdesk.

The final advantage was that CSIRO could easily identify users from those external groups. This turned out to be the underpinning of the process that would give them access to the publishers titles to which they were entitled to see.

Storage
One area that did require rethinking was storage requirements. An estimate of the amount of storage required for all the full-text articles was 200MB per title per year. The loading of 1200 titles back to 1995 would soon see requirements in excess of 1TB. A Storage Architecture Network (SAN) solution from Storage Tek was adopted by CSIRO.

The advantage of the SAN storage environment is that it allows all holdings to be available on-line.

A summary of the storage configuration at December 2000 is:

• 1.2TB online storage
• Robotic (DLT) tape backup
• All holdings are available online
• Architecture provides high availability and high levels of redundancy
• Used for other corporate applications also

Subsequent progress

CSIRO's goal is to aggregate and integrate as much electronic journal content into the robust infrastructure that has been developed. At December 2000 CSIRO had 1.3 million journal articles fully indexed and searchable online from all locations. They include:

• Elsevier (1200 titles from 1995)
• CSIRO Publishing (13 titles from 1997)

Other full text publishers likely to be added in early 2001 include: Kluwer, Springer, and Blackwell Science. CSIRO has initiated further negotiations with other publishers and is seeking to obtain critical mass through depth and breadth of title coverage. A key consideration is making the collection as rounded as possible to ensure that there is something of value to all scientists regardless of discipline.

The ScienceServer software can also support the loading of secondary databases. The Compendex database covering 1970 onwards went live in late 2000. Trial loading for other secondary abstracts including Current Contents and the SCANFILE internal database has been successful. An advantage of loading both full text and secondary databases is that greater integration and linking is possible.

Linkages

Links have been developed in and out of ScienceServer to other sources:

• Web of Science to full text
• Chemical Abstracts Services' SciFinder to full text
• Bi-directional links to CSIRO library catalogue.

The ScienceServer application handles articles in Portable Document Format (PDF) and it is planned to manage HTML/XML formats in the future. While many early issues are scanned with variable quality, 95 percent of articles are now distilled PDF and of much better quality.

CSIRO is technically unable to load HTML article files into ScienceServer at present. In addition a number of publishers with whom CSIRO is negotiating will not permit local loading for technical reasons or market positioning strategies. However it is possible to integrate content from publishers' sites through incorporation of metadata into ScienceServer. Effectively this allows the user to search locally loaded and external content through a single front end. If non-locally loaded full text is required, a 'pass through' is facilitated to the publishers' URL.

This is the solution being pursued at present with:

• Wiley Interscience (315 titles, the majority dating from 1997)
• American Chemical Society (27 titles dating from 1996)

Usage

The uptake of electronic products within CSIRO has been particularly encouraging. Although there were concerns that scientists preferred print format, they have embraced the technology with minimal fuss. In the annual CSIRO Library Network User Survey in 2000, 86% of respondents considered CSIRO's speed in introducing electronic resources was 'about right', while a further 12% consider the speed to be too slow.

When new resources are announced across CSIRO, positive e-mail replies are received together with shopping lists for additional titles that are sought. One chief of a research division commented that he had recently written a paper in an afternoon. He compared this with two weeks of searching for articles in the print environment! Others have commented that CSIRO's competitiveness is enhanced through our ability to attract new staff through the array of titles available to support research.

One of the biggest gains has been in the access from small and remote research stations. Staff in these areas have equal access with their capital city colleagues. Another plus has been the revival of access to some expensive titles that had been cancelled in recent years. Savings have been achieved through reduced circulation of print journals across the Organisation and some sites are starting to see advantages in reduced demand for shelf space!

A web front end to interrogate usage statistics has been developed but further work is still required. One positive will be consistent data for all publishers loaded into the system compared with the reliance on and variations in usage data between publishers with whom we obtain journals via the internet.

Tom Sanville from OhioLINK has published detailed usage data. Of the 400,000 article downloads through June 1999, he notes that 51% were from journals not held in print at the downloading patron's library. Of the 1300 titles loaded at CSIRO in January 2001, only 48 titles have not been accessed. The majority of these are in foreign language or non-science disciplines.

The library community is polarised over the benefit of purchasing entire publisher collections. Some suggest that the benefit of the additional titles is questionable. On the other hand, Sanville (1999) suggests that:

The results thus far do indicate that librarians should no longer presume to know exactly what patrons will need in the electronic world based solely on past patron behaviors in the controlled print environment. We need to avoid electronic solutions that are based on the old rules of limited information use and single site economics. Rather, we should seek solutions that maximise our ability to let information use expand and seek new levels.

Single user interface

The concept of a single user interface has been the 'holy grail' for many organisations for some time. The plethora of publisher sites and their different interfaces and search engines creates an added layer of complexity for the user, although this is likely to diminish over time as systems become simpler and the user base more sophisticated.

Through local loading into the EJC, CSIRO is following an 'interface culling' strategy. Efficiencies in operations, administration and training accrue as a result. However it would not be viable or possible to load all electronic journals locally.

In addition, scientists access information through diverse pathways. The initial entry point is often minor in comparison with the variety of systems through which users navigate. For example, the interface of choice for scientists may be abstracting and indexing services such as Web of Science. Linking from article references automates an age old information seeking activity. While CSIRO will continue to seek administrative economies in storage and software systems, information diversity must be supported through the hybrid library approach.

Conclusion

The CSIRO EJC has been a very successful project. The technology choices and the local loading solution are suitable to CSIRO and its infrastructure. The ScienceServer community of users has a similar profile and most sites are characterised by strong leadership and an innovative approach. In CSIRO's case this project was lubricated by the structural relationship of corporate library and information technology staff. It is unlikely that small organisations or libraries would pursue this path.

Many ScienceServer sites worldwide have a consortium or national perspective (eg OhioLINK, Taiwan). Sites such as Los Alamos and the University of Toronto provide hosting facilities to external organisations (eg other government science organisations or neighbouring universities).

CSIRO has initiated discussions with other Australian commonwealth science agencies and universities with a view to hosting ejournals on their behalf. Discussions with a number of international publishers has centred on the possibility of CSIRO providing local archive or mirror facilities. Should additional government funds be made available for a national digital science resource the CSIRO installation could logically become a service node.

In addition to the many potential opportunities that may be explored, CSIRO's priorities for the future include:

• Aggregating additional content from other publishers, and selected earlier issues from existing publishers.
• Integrating further with other sources and citation linking. (eg. linking to citing articles).
• Improving and streamlining existing procedures for content delivery from publishers.
• Working with ScienceServer LLC and other users to enhance the system, its functionality and usability.

In April 2000, Elsevier Science announced their merger with Endeavor Information Systems. Interestingly CSIRO was an early adopter of both ScienceServer and Endeavor's Voyager systems. The effects of this merger have yet to be realised but CSIRO is well positioned and has flagged an interest in becoming a beta site for any software outcomes.

The current solution is appropriate for CSIRO and should serve the Organisation effectively for another 3-5 years. New forms of scholarly communication will develop into the future and systems must be scalable to accommodate new functionality and requirements.

Technology has been pivotal to this project. However the most important outcome has been the speedy delivery of enhanced access to scientific literature to support CSIRO's research enterprise.

References

Pack, Thomas and Pemberton, Jeff 'A Harbinger of Change: The Cutting Edge at the Los Alamos National Laboratory' ONLINE March 1999, pp 34-42

Priorities and Pathways: Review of the CSIRO Library Network (1997) Canberra: CSIRO, Internal confidential report

Sanville, Tom 'Use Levels And New Models For Consortial Purchasing Of Electronic Journals' Library Consortium Management: an international journal V1 No 3/4 pp47-58, 1999

Websites

Australian Nuclear Science and Technology Organisation (ANSTO) http://www.ansto.gov.au/

Commonwealth Scientific and Industrial Research Organisation (CSIRO) http://www.csiro.au

Los Alamos National Laboratory Research Library and Library Without Walls http://lib-www.lanl.gov/fancy-index.htm

Ohio Library and Information Network (OhioLINK) http://www.ohiolink.edu

ScienceServer LLC http://www.scienceserver.com/index.html

University of Toronto Library http://www.library.utoronto.ca/