ROADS offers a solution, free at the point of use, in an area with very few commercial alternatives offering comparable or appropriate functionality and flexibility to information gateways.

ROADS partners are active in World Wide Web Consortium (W3C) working groups, those of the Internet Engineering Task Force (IETF), the Trans-European Research and Education Networking Association (TERENA) and other UK and international consensus-making bodies at technical, service and organisational levels. The standards-based ROADS tools, methods and protocols which result from this close involvement enable subject specialists to build and maintain information gateways (collections of selected high quality Internet resources) and end-users to make complex, transparent searches across distributed information gateways.

2. Activities and progress against objectives

ROADS has continued to fulfil the project objectives and has made increasing contributions to the ongoing success of the ANR services and others. The activities and progress over the reporting period are summarised here and are grouped broadly by each of these original objectives.

2.1 Sharable, distributed systems

The resource discovery systems that have been designed and implemented to date by ROADS continue to support and extend the core functions of a majority of the ANR services. Much of the tools development and liaison work which has occurred over the last year has concentrated on consolidating and extending support for these services, as well as introducing new functionality developed in response to the changing needs of these services and developments in related technologies and standards.

2.2 Support for subject-based services

2.2.1 Hybrid Libraries

The ROADS project appreciates the importance of the development, this year, of the Hybrid Libraries strand of the eLib Phase 3 programme and a meeting was held on the 21^st of May 1998 that provided an opportunity for representatives of HeadLine (an eLib Hybrid Library project) to exchange ideas with the ROADS team and other ANR service staff. The interoperability of ROADS-based services means that there are many potential uses for ROADS within the broader scope of these projects.

A member of the ROADS team has also met with representatives of the BUILDER Hybrid Library project at Birmingham University, who, like the HeadLine project are interested in incorporating elements of ROADS tool-kits into their systems.

The development of a `shrink wrapped' Z39.50 add-on module for ROADS (see below) has facilitated closer participation between ROADS gateways and the eLib Phase 3 projects. Biz/ed and SOSIG, for example, have been participating in the Agora programme through provision of experimental Z39.50 interfaces based on this technology.

Various members of the ROADS team have also attended related eLib 3 meetings.

2.2.2 Take-up and use of ROADS software

The take-up of ROADS tools for the provision of subject-based gateways and other types of services has grown enormously over the past year. Much of this growth is due to the efforts of the ROADS liaison officer and the involvement of the ROADS project within the EU-funded DESIRE initiative. Participation in the DESIRE project has meant a number of innovations for ROADS, including the new multi-lingual functionality included in version 2.

The following ANR projects have continued to manage and develop ROADS-based information gateways:

ADAM <URL: http://www.adam.ac.uk/ >

Biz/ed <URL: http://www.bized.ac.uk/ >

HISTORY <URL: http://www.ihrinfo.ac.uk/ >

OMNI <URL: http://www.omni.ac.uk/ >

SOSIG <URL: http://www.sosig.ac.uk/ >

Non-ANR gateways using ROADS include (but are not limited to ):

ALEX <URL: http://sunsite.berkeley.edu/alex/ >

A catalogue of electronic texts. The gateway index contains some 2,000 resources, subject matter being the full text of publications e.g. "War and Peace" in its entirety.

The Countryside Recreation Network <URL: http://localhost/crn/>

Database of references to publications produced in 1995 by the CRN agencies.

EELS <URL: http://www.ub2.lu.se/eel/eelhome.html >

A quality controlled subject service, covering the broad subject area of Electronic Engineering.

NOVAGate, the Nordic Gateway to Information in Forestry, Veterinary and Agricultural Sciences <URL: http://novagate.nova-university.org/>

Further information about these and other ROADS based services are available from:

<URL: /who/ >

ROADS is also being used in several UK HE institutions to provide locally accessible information services. Noteworthy examples are the Open University and the Edward Boyle Library, University of Leeds. The Finnish Virtual Library project have also committed to the wide scale use of ROADS.

The Social Science Grapevine, a jobs and training gateway for the social sciences, <URL: http://www.grapevine.bris.ac.uk/ > is also using components of the ROADS system to interface with a relational database storage back-end, whilst providing standards based cross-searching support.

This means that there are over fifteen public ROADS-based services currently operating. Over the past year there have also been 256 downloads of various ROADS software releases, and 87 registered ROADS installations. (New users are prompted to register their system during the installation process). There have been 79 downloads of ROADS version 2 thus far.

2.2.3 New development models

The increase in the size of the user base combined with the open nature of ROADS software development has meant that with each new installation the number of potential ROADS developers has grown. This has led to the development of code by some members of the user community, further distributing the ROADS coding effort. New development by these `ROADS hackers' are immediately fed back into the software iteration cycle to improve the code base.

Examples of this range from suggested version 2 beta bug fixes provided by Bill Jupp of the Edward Boyle Library, University of Leeds to whole new areas of functionality, such as a utility for the automatic generation of PDF files from text documents to aid readability when printed. This was produced by Eric Lease Morgan, administrator of the Alex service.

This approach is facilitated by using the roads-hackers mailing list. Increased take-up has meant that the project has matured more quickly as a result of this extra development work.

It is clear that this is the way forward as far as ROADS software development is concerned and the code base is being actively modified and modularised to make this process easier. An example of this is the development of the ROADS API, detailed below.

2.2.4 Software releases

The ROADS project continues to apply strict version control over software releases. Information about the current release status is available from the Web site. This is to ensure that ANR services are both kept abreast of the latest developments and also that they are clear which software versions are suitable for implementation on live services.

The second year of the ROADS project saw the release of ROADS version 1. During implementation particularly on the SOSIG subject gateway a number of bugs were uncovered, as is usual. Patches were quickly developed and released but this meant installation for new services was more time consuming.

In order to rectify this the ROADS v1.01 distribution was released in January 1998 to avoid users needing to apply a number of patches. The current stable release is now ROADS v1.02. This is a maintenance release of ROADS version 1. The ROADS version 1 code has been frozen and will not be altered now except for bug fixing.

This year ROADS version 2 went into the beta test phase and at time of writing is currently available as version 2 beta 3.

This testing is being done by the Social Science Information Gateway (SOSIG) and Biz/ed in Bristol. Other ROADS-based gateways have also participated in the beta testing programme; including the WoPEc gateway to economics working papers.

This new version addresses issues highlighted during the `version 2 requirement gathering process' as detailed in last year's report, and is intended to keep the development of ROADS in step with the needs of the ANR services. The new functionality has been added to ROADS version 2 as follows:

This last feature extends the forward-knowledge querying approach to the Z39.50 protocol which will form a part of systems developed under the Hybrid Libraries strand of eLib phase 3. This is highly significant for any future integration and interoperability between Hybrid Libraries and ANR services.

2.2.5 Improving the ROADS installation and configuration process

One of the issues that was raised from various user feedback sessions undertaken was that the initial installation process left room for improvement. Installing a ROADS version 2 gateway has now been made considerably easier. The new installation process includes a `tips' option which makes suggestions about configuration choices (this uses a set of heuristics which examine your UNIX system), and takes you step by step through setting-up and configuring your new ROADS gateway.

To improve this process still further Bristol have put the new installation routine through `quality assurance' tests, including the detailed observation of installations conducted by technically astute staff, previously unfamiliar with ROADS. Several amendments have been made as a result of this.

We hope to extend this process to other aspects of the software in order to complement our alpha and beta testing programmes.

2.2.6 Year 2000 compliance testing

ROADS version 2 (beta 1) has been tested for year 2000 (Y2K) compliance, using the British Standards Institute checklist by the team at Loughborough. The results of this testing can be found at:

<URL: /papers/y2k/ >

No problems were discovered.

2.2.7 ROADS version 3

The ROADS version 3 code base is now well under way and will soon be ready for alpha release. During the development of ROADS it has become clear that much of the functionality we are producing for eLib subject gateways would be more generally useful to other developers, also it is clear that the ROADS back-end database is not always the most convenient way to integrate ROADS functionality into other types of services.

2.2.8 ROADS version 3 API

To this end, it was decided that ROADS version 3 should clearly separate the various portions of commonly used code that would be of use to developers of other networked information discovery and retrieval applications. This involved developing a clearly defined Application Programming Interface (API) between the ROADS software and other third party applications. The API would both allow developers to add support for their third party code to ROADS without having to modify the ROADS code base itself, and provide a mechanism for the use of ROADS for hitherto unsupported / unforeseen activities.

Initial requirements for version 3 were gathered at a meeting attended by representatives of ROADS-based ANR subject gateways. We originally intended to hold a workshop with more ROADS users, to gather feedback on the features to be included in the ROADS API. In practice it proved difficult to agree a date when all interested parties could meet. We decided to experiment with tele-conferencing instead to see if it would be possible to hold a `virtual workshop'. We decided to use Internet Relay Chat (IRC), and managed to run a very successful two day workshop during which the details of the API development were decided.

The workshop agenda, transcripts, and an executive summary are available at:

<URL: /papers/ >

The API has been implemented for ROADS version 3 alpha. The costs involved in using IRC for the meetings is especially interesting. A workshop in London would have cost around £1,000 whereas the IRC sessions cost the project nothing to hold. In addition to saving money we observed a number of benefits from this approach - e.g. increased participation (projects did not have to delegate a member to attend on their behalf), the ability to refer back to online technical material whilst `in' meeting, and the availability of a full-text transcript, generated automatically, which obviated the need for minute taking.

Because of this success we will be using IRC for future meetings where appropriate.

2.2.9 Improvements for ROADS cataloguers

A cataloguers workshop was held on the 18^th of November 1997 attended by representatives from several ANR services and other ROADS users. Comments and suggestions were gathered concerning the ROADS version 1 Web based template editor. Much of these comments centred around making the editing process more simple, with easier navigation around the various editing forms. These comments are summarised on the Web site:

<URL: http://www.roads.lut.ac.uk/v3/v3-req.html >

In response to these, many of the requested features have actually been included in the new interface for version 2. However, as part of the version 3 requirement it was decided that a range of options for cataloguing would be more helpful for users rather than trying to squeeze all the requested functionality into the Web based forms, which would have involved a number of compromises.

Therefore, in addition to the improved Web based interface in version 2, Jon Knight is now developing a stand alone template editor for version 3. This will be run on the end-user's desktop rather than via the Web and utilises the new ROADS API. This should make cataloguing faster, allow new features to be added more easily and make record creation / editing a more pleasurable experience.

The new editor is coded using the Perl/Tk graphical user interface scripting language and is intended to run under either X-Windows or Win32 (95/98/NT). A similar utility can also be produced for Macintosh operating systems if there is a requirement from the user community. Other techniques for cataloguing involving the use of Microsoft Access are also being investigated. The aim is to provide ROADS users with a range of techniques for populating their databases.

These developments were specifically requested by ROADS users.

2.2.10 Modularising the ROADS code base

A parallel goal in version 3 has been to reduce the interdependency between sections of the ROADS library code. By doing this, it is hoped we will arrive at generic reusable library. Our intention is to post individual library modules on the Comprehensive Perl Archive Network (CPAN), <URL: http://www.perl.com/CPAN/ > a globally distributed repository of usercontributed Perl library modules and programs.

The aim in so doing is to move ROADS development onto `Internet time' rather than the artificial process of monthly alpha and beta releases. This means that ROADS code will be used more widely, and developed, improved, added to and be made more robust, more quickly.

This is essentially following the development model that has been so successful for the Linux operating system and which Netscape have now adopted by releasing the source code of their Navigator browser under the Mozilla.Org initiative.

In addition to posting our library code on CPAN, we have also instituted a number of measures to make the ROADS code base more accessible to developers:

2.2.11 User interface testing

This year SOSIG is undertaking user testing of the interface to their gateway. The first session took place with a group of students from the School of Policy Studies at Bristol. The aim of these sessions is to determine how best to integrate new features such as a Harvester service and an online thesaurus. It is intended that findings from these sessions will be fed back into the ROADS design cycle so that future iterations of the ROADS interface can include these new features.

2.2.12 Template Registry work

The project has recognised that ROADS templates, in common with other metadata formats specifically developed for electronic resources, need to adopt some of the characteristics of traditional cataloguing standards. This has resulted in the creation of a metadata registry for ROADS templates, which is based at UKOLN. This registry lists and briefly defines all template-types and data-elements used by ROADS-based services. The existence of this registry means that there need be no unnecessary proliferation of service-specific template-types or data-elements, thus helping to ensure the on-going interoperability between different ROADS-based services.

<URL: http://www.ukoln.ac.uk/metadata/roads/templates/ >

2.2.13 RECCI

Now ROADS services are capable of being cross-searched it is important that there is at least some semantically consistent use of ROADS templates. In order to investigate these issues, and to back up the work of the Template Registry and the production of some generic ROADS Cataloguing Guidelines, a study of the current cataloguing practices of subject services is being prepared. This study is being carried out by UKOLN and is known as ROADS Evaluation of Cataloguing with Connection to Interoperability (RECCI). It includes an analysis of current cataloguing policies and practice and their impacts on interoperability, together with information on services' editorial policies and a statistical study of template use. A sample of ROADS templates have been collected from eLib subject services (ADAM, History, OMNI and SOSIG) and the contents have been evaluated with regard to quality and cross-searching ability.

<URL: http://www.ukoln.ac.uk/metadata/roads/recci/ >

2.2.14 Cataloguing Guidelines

Interoperability and ROADS cross-searching will depend upon the existence of some semantic commonality between subject services. Within ROADS, this semantic commonality is best served by the creation of generic cataloguing guidelines. Accordingly, in September 1997 a ROADS cataloguing guidelines meeting was held in Bath between representatives of the eLib ANR services and ROADS personnel. UKOLN produced a first draft of some generic cataloguing guidelines in January 1998. These provided guidance on formats and codes that could be used for things like names, dates and languages. A revised edition of the rules was produced in July 1998. In addition to their potential usefulness in a ROADS-based service context, the development of cataloguing standards for Internet resources may help to promote improved semantic interoperability between metadata formats.

<URL: http://www.ukoln.ac.uk/roads/cataloguing/ >

2.3 Support for information providers

2.3.1 ROADS databases of metadata for Web site administration

UKOLN have been investigating the use of ROADS to hold metadata for local Web site administration. Records describing local resources are stored in a ROADS database and then embedded into HTML META tags using a Server Side Include (SSI) script as the pages are served.

The system makes use of the proposed DUBLINCORE ROADS template type also being developed at UKOLN.

The system might then be used in the following way:

This means that ROADS can be used to generate metadata by the information providers themselves. It would also then be possible for subject gateways to extract this metadata from the administration databases for inclusion in the gateway catalogue, saving cataloguing effort.

2.3.2 Contributions to the DESIRE project

DESIRE was one of the largest projects funded in the Telematics for Research Sector of the Fourth Framework Programme funded by the European Union. The partners in DESIRE have been at work since January 1996 on a series of related tasks which extend the technology of the World Wide Web and implement pilot information services on behalf of European researchers.

Work package 3 of DESIRE was particularly relevant to ROADS and subject gateways. This work package was aimed at meeting the needs of research users for better means of locating the information that is relevant to their research endeavour. The package contained two major elements: subject-based information gateways (distributed resource discovery based on rich descriptions and quality-controlled resource catalogues based on specific subject areas) and exhaustive automated indexing of existing World Wide Web information sources in all subject areas. Verifiable objectives were:

User groups for verification included Social Scientists (University of Bristol), Engineers (Lund University) and Art Historians (Koninklijke Bibliotheek).

As part of the DESIRE project a training work package was put together with the SOSIG (the Social Science Information Gateway) team and Netskills, for information gateways based on distributed cataloguing models. These materials were delivered at a workshop in December 1997 to a group of staff from ANR services (including OMNI, ADAM, HISTORY (IHR-Info), NETEG, and EEVL) also present were other representatives of other ROADS services from Belfast, Sweden and the Netherlands. The workshop was aimed at those providing training to cataloguing staff for their ROADS-based services. The latest developments in ROADS tools were also showcased.

The work within the DESIRE programme has also resulted in the development of multi-lingual functions within the ROADS tool-kit.

2.4 Standardisation

From its instigation ROADS has been committed to making its tool set interoperable both among ROADS-based services and services running other protocols such as Z39.50.

2.4.1 Forward-knowledge based querying in ROADS version 2

As part of this interoperability initiative, ROADS has been developing facilities to allow centroid gathering and query routing using RFC 1913 mesh traversal.

This means that ROADS provides support for `forward knowledge' based query routing. Collaborating services can periodically extract a `centroid' from each ROADS server, which lists all the unique words in every field of the database.

This means that users can `cross search' multiple ROADS services with a single query and have meaningful results returned. The benefit of using the forward-knowledge approach is that the system consults the centroid of each service before submitting the query, ensuring that searches are not broadcast indiscriminately to multiple servers: a query is only sent when there is some likelihood of a successful search. This is designed to scale well to resource discovery scenarios in which users may want to simultaneously query a large number of targets. In the context of bandwidth charging and international collaboration, it is important to make the most efficient use of network resources when designing large scale resource discovery systems. The Loughborough ROADS team's participation in the IETF ASID working group is a part of this work. ROADS version 2 (currently in beta test) provides enhanced support for cross-searching and forward-knowledge based query routing.

2.4.2 Cross Searching demonstrator (CrossROADS)

A demonstration of the centroid-based cross-searching functionality in ROADS version 2 is now available. Searches using this demonstrator will be processed by an index server which periodically visits several of the eLib subject services as well as other gateways like the ROADS-based ALEX electronic texts gateway and the EELS Engineering Electronic Library, and extracts a summary (centroid) of their databases. A search through the index server will determine which of the subject gateways will have information that matches the query. If requested, the query can be passed on to each subject gateway with matching database entries and these can then be retrieved for display to the end user. This is a vast improvement on the prior practice of conducting multiple parallel or sequential searches of the servers in a `cluster'.

It also provides one of the earliest implementations of the Common Indexing Protocol Version 3 (CIP v3) which is an IETF standards track enhancement to WHOIS++ style centroids. This work allowed a number of suggestions to be fed back from the ROADS team to the IETF FIND working group in order to clarify and fix problems in the draft specification.

Note that although EEVL is not a ROADS-based service they have developed a centroid mechanism which allows interoperability between their proprietary software and ROADS.

<URL: http://www.ukoln.ac.uk/metadata/roads/crossroads/ >

There are issues of interface design as well as problems of duplication and declaration of source in results reporting which arise out of this work. Investigation of these issues is ongoing.

This work was published in a paper in the January 1998 issue of D-lib entitled Cross Searching Subject-Gateways: The Query Routing and Forward Knowledge Approach produced by members of the ROADS team and Sue Welsh of OMNI.

<URL: http://mirrored.ukoln.ac.uk/lis-journals/dlib/dlib/dlib/january98/01kirriemuir.html>

2.4.3 Cross-searching and forward-knowledge querying with SWISH-E

Also as part of the ILRT contribution to the ROADS cross-searching architecture, Dan Brickley has been working on tools that will make it possible to cross-search ROADS databases with HTML-based Web site indexes. This consists of a set of extensions to the SWISH-E searching and indexing tool. SWISH-E is already used by many sites as a fast, efficient, and freely available tool for creating metadata aware single-site search facilities.

A WHOIS++ `wrapper' is added to a standard SWISH-E installation, making any SWISH-E site index accessible to WHOIS++ search clients. This means that it is now possible to search, from a single interface, any combination of ROADS databases and SWISH-E Web site indexes.

A centroid-generator is included in the package, allowing ROADS based services to share forward-knowledge about search terms likely to produce hits on collaborating Web servers. This means that SWISH-E indexed sites will be able to take part in a cross-searching mesh without being bombarded with searches on their server. Centroid aware search agents will filter searches first and only send queries likely to produce hits.

SWISH-E extensions are planned to ship alongside version 2 of ROADS.

2.4.4 NewsAgent and ROADS

NewsAgent is an eLib project developing a current awareness service for LIS based on an Oracle database developed by Fretwell Downing Informatics. The database currently contains around 2500 resource descriptions. UKOLN have developed a WHOIS++ front-end to the NewsAgent database using the WHOIS++ daemon supplied with ROADS. It supports the generation of centroids. Work ongoing at UKOLN includes incorporating NewsAgent into the growing mesh of cross-searchable ROADS-based services.

<URL: http://roads.ukoln.ac.uk/newsagent/cgi-bin/search.pl >

2.4.5 ROADS Harvesters

ROADS has provided an experimental HARVEST-based 'combine-harvester' add-on for the ROADS software. This enables:

<URL: http://www.ukoln.ac.uk/metadata/roads/harvester/v1a1/ >

2.4.6 The SOSIG Harvest demonstrator

Members of the SOSIG team have produced a demonstrator using harvesting technology that has been developed as part of the DESIRE project. This has indexed the top two levels of each Web site described in the SOSIG catalogue.

This has then been used to create a ROADS database of some 40,000 document templates. This means that users would be able to search either the metadata contained in the SOSIG catalogue or perform a free text search of the resources described.

An unexpected benefit of this work has been that we have demonstrated that ROADS is scaleable for use with datasets containing tens of thousands of records. Further scaleability testing will be performed later this year.

<URL: http://edward.ilrt.bris.ac.uk/combined-roads/cgi-bin/search.pl>

Again, SOSIG are looking at the interface design issues arising out of combining this type of functionality with the existing service. This work will be fed back into the ROADS design cycle.

2.4.7 Z39.50 and WHOIS++ interoperability

Resource descriptions held in a ROADS database are normally made available via WHOIS++, a simple, yet powerful Internet search protocol. However, it is clear it is also desirable to make ROADS databases available to end-user clients (and intermediate systems) that use the Z39.50 search and retrieval protocol, even though there are a number of drawbacks in using this protocol for ANR services.

The ROADS project has achieved this interoperability with the development of a Z39.50 to WHOIS++ gateway called ZEXI. This gateway functions as a Z39.50 server, accepting queries from Z39.50 client systems. It then converts them into WHOIS++ queries and passes them on to a ROADS server. As the ROADS server returns results, they are converted into a suitable format for use by Z39.50 client systems and returned to the client as a Z39.50 result set.

ZEXI is based on the Isite Information System available from CNIDR and returns simple, unstructured text-based records known as SUTRS. ROADS (UKOLN) is in the process of enhancing this demonstrator to enable it to return structured USMARC or GRS-1 records to the Z39.50 client.

An alternative approach to ZEXI involves copying records from a ROADS database into some other database that has a Z39.50 interface. For example, the Zebra Z39.50 server can make converted ROADS records available in two structured formats (USMARC and GRS-1) and in an unstructured format (SUTRS). UKOLN have also developed a script <roads2gils.pl> for converting ROADS templates into an SGML-like GILS record suitable for importing into Zebra. The eLib ANR subject services SOSIG and Biz/ed will be participating in the Agora Hybrid Library (eLib Phase 3) project in this latter way.

<URL: http://www.ukoln.ac.uk/metadata/roads/interoperability/roads-z3950.html >

This year Loughborough have also developed a tool that allows a Z39.50 server to be queried with a WHOIS++ client. Although still very much in alpha release this means ROADS systems can now be both searched using Z39.50 clients and query Z39.50 servers. This uses code contributed by Peter Valkenburg of TERENA as part of the development of the CIPv3 standard detailed below.

ROADS is moving towards a position where it should become trivial to add Z39.50 capabilities to any standard ROADS installation. A highly significant development for the future of ANR services.

2.4.8 Dublin Core Templates

ROADS partners have continued their close involvement in the Dublin Core (DC) initiative. Project partners from Bristol and UKOLN attended the 5th Dublin Core Metadata Workshop on the 6-8 October 1997 in Helsinki, Finland. In addition, Lorcan Dempsey (UKOLN) sits on the DC Policy Advisory Committee and Rachel Heery and Andy Powell (UKOLN) are members of the DC Technical Advisory Committee.

Work has been carried-out at UKOLN into how a ROADS database can be used to manage Dublin Core metadata across a Web site. Records describing local resources can be stored in a ROADS database and then a Server Side Include (SSI) script can embed the metadata in HTML <META> tags when the relevant page is served. Accordingly a DUBLINCORE template type has been proposed which would contain all 15 DC elements, the sub-elements proposed by DC sub-groups after the DC-5 meeting in Helsinki and associated SCHEME qualifiers. The SSI script reads the ROADS DUBLINCORE template and converts the metadata into HTML <META> tags or (experimentally) into an XML representation of Resource Description Framework (RDF).

<URL: http://www.ukoln.ac.uk/metadata/roads/metadata-mgmt/ >

2.4.9 The TF-CHIC pilot project

In an outgrowth of the work on indexing done for the DESIRE initiative, ROADS software is now being used on an ambitious project to build a distributed searchable index of European academic WWW pages, under the auspices of the Trans-European Research and Education Networking Association (TERENA).

The goal is to build upon existing funded and volunteer based WWW indexing efforts in Europe (examples include AC/DC in the UK <URL: http://acdc.hensa.ac.uk/>), using WHOIS++ as a common search and retrieval protocol, and investigate the feasibility of using centroid technology for distributed creation and searching of WWW indexes. This work integrates diverse products from commercial vendors (e.g. AltaVista and InfoSeek) in addition to free software products such as Harvest and the Zebra Z39.50 server.

This work has given the ROADS software the ability to interact with Harvest and Z39.50 servers at the WHOIS++ protocol level as though they were WHOIS++ servers. Included has been support for the exporting of centroids for use in an index server.

<URL: http://www.terena.nl/projects/chic-pilot/ >

2.5 Dissemination

Dissemination is particularly important to a project such as ROADS, where our main output is an actual product. Dissemination both assists and encourages take-up, and keeps existing users up to date with the latest developments.

2.5.1 ROADS Web presence

The ROADS Web site <URL: / > has been significantly expanded, both in terms of content, and in terms of visibility, over the last year. Sections of the Web site have been allocated to the three project partners; these sections are seamlessly inter-linked by a common `house style'.

The style of the Web site, deliberately matches that of the ROADS flyer, giving a consistent look and feel to ROADS publicity and information sources. The Web site is split between the three partners so that credit is shared equally and the profile of each partner is the same.

The ROADS Web site contains twelve sections. These are allocated to project partners who work most closely on the associated tasks. It should be noted that some sections, and their associated tasks, involve more than one partner; for example, all three partners work on ROADS interoperability issues.

By splitting the ROADS Web pages across three sites we offer an increased number of links and entry points to the pages. In the next year of the project, some evaluation and investigation will be undertaken of who uses the ROADS software and how they become aware of the system.

The twelve sections of the Web pages are as follows:

Some areas of the Web site hold limited content at present. This is because we have designed the site to allow for future expansion without time consuming redesign.

We anticipate the need to accommodate certain types of reports, guides, software tools and other materials over the coming years, and have designed the site accordingly. For example, as more related initiatives progress, we expect to add relevant details about them to the section on related initiatives.

2.5.2 ROADS Newsletter

The ROADS Newsletter is a Web only publication. Prior to January 1998 the Newsletter appeared on an irregular basis to announce key milestones in the progress of the project. At the beginning of this year it was decided that the Newsletter would become a regular quarterly publication to reflect the growth in activity of the project and to cater to the expanding user base. Editions have since appeared in April and July.

The newsletter has expanded to include not only material about ROADS, but information and opinion on associated developments and initiatives. Every issue now contains core material on:

The ROADS project recognises that there are many resource discovery and gateway developments outside of the scope of our remit. Because of this, and to position ROADS more clearly in the context of other developments, we have asked various other services, organisations and people to write for the newsletter. This also means that our newsletter has a wider audience than if it were just restricted to ROADS-based services and developments.

Recent authors from non ROADS-based services have included BUBL, EEVL and Telerise. A regular columnist, Charlotte Jenkins, writes on matters concerning resource discovery; in addition, we now review at least one high-profile, non ROADS-based gateway in every issue, partially to demonstrate our acknowledgement of the diverse range of such systems available to the public. A current awareness section in each issue also links to various resource discovery articles that can be found on the Web in journals such as Ariadne and D-lib.

The newsletter is widely publicised on relevant mailing lists in the UK, the US and Europe. Web statistics gathered from the Bristol section of the ROADS pages indicate that people from a diverse range of countries and organisations read articles within the newsletter. More formal evaluation needs to be undertaken, but it appears that the regular appearance of the Newsletter has improved its effectiveness as a means of dissemination.

2.5.3 ROADS flyer

Towards the end of 1997, a flyer for the ROADS project was created. This consists of an overview of the projects services, partitioned into the categories of Support, Standards and Software. The flyer contains screen shots of various ROADS-based services, as well as of some of the ROADS functions in operation. The flyer includes all of the project contact details and logos of the nine public services who were using ROADS at that point.

2000 of the flyers were printed. Bundles of these were despatched to all ROADS-based subject gateways for cross-publicity purposes. Bundles of flyers have also been despatched to a wide range of other organisations and interested parties on request. These have included Hybrid Library projects such as HeadLine and BUILDER, various JISC and eLib support units, visitors to the ILRT, and many CTI centres. Flyers have also been distributed at relevant conferences and training workshops.

2.5.4 Mailing list publicity

Periodic postings to subject gateway relevant and resource discovery mailing lists are made when appropriate, notifying people about ROADS. These include many of the major mailing lists that cover the wider field of resource discovery, especially those that have a significant number of technically-oriented LIS staff as list members. For example, Web4Lib, a large US-centric list, is targeted for posting as it has a significant number of potential gateway creators.

Mailing lists were identified, as part of the ROADS dissemination strategy, by searching list indexes such as Mailbase, Tile.Net, Liszt, and the Kovacs list archive. It is estimated that ROADS has been either mentioned, or advertised, on between 130 and 150 mailing lists in the last year. Most of these were overseas mailing lists, as ROADS has been keen to `spawn' gateways overseas. This policy has proved successful, with people such as Joseph Edwards (Waste Water Engineering Virtual Library) and Eric Lease Morgan (ALEX) creating quality ROADS-based gateways, as well as contributing ideas, comments and code into the ROADS effort. In addition, some of the gateways being established, such as the Law gateway at PACE University and the AALN (American Arts and Letters Network) may prove to be strong candidates for cross-searching with ANR services that have overlapping subject areas, such as SOSIG (subsection on Law) and ADAM (Arts).

2.5.5 ROADS Liaison Activity

The ROADS Liaison Officer has made visits to a number of ANR services using ROADS over the past year in order to keep up to date with service developments and user requirements. Demonstrations have also been arranged with other ANR services and interested parties throughout the year. Visits have been made to JILT, OMNI, ADAM, HISTORY (IHR-Info), the British Library, and a number of key users in Finland and Holland have. A very fruitful meeting between HeadLine and the ROADS team was also organised.

Visits have been made to a large number of organisations who have expressed an interest in using ROADS, including the Cranfield Institute and the National Maritime Museum. There are a number of proposals for ROADS gateways and services in the pipeline arising from these contacts.

2.5.6 The ROADS related mailing list

The ROADS team at Loughborough maintains a number of electronic mailing lists both directly and indirectly related to the ROADS project, each focused on a slightly different target audience. All the lists are archived on the Web and available at:

<URL: http://www.roads.lut.ac.uk/lists/ >

Loughborough also hosts `meta2' - the main mailing list of the Dublin Core metadata initiative.

2.6 Outputs

2.6.1 ROADS releases

A number of maintenance releases of ROADS version 1 have been made during the year. ROADS version 1 has proven to be a stable release for production services.

ROADS version 2 has gone through a number of alpha and beta releases and is on the verge of entering its full production release at the time of writing.

In the last few months, ROADS version 3 alpha code has also been released. The overlapping of releases allows a more rapid turn around and helps to eliminate `dead time' where the developers are waiting for bug reports before coding the next.

Version 3 timeline (template editor & API alpha releases)

v3a2 - June 26 1998

v3a1 - June 10 1998

Version 2 timeline

v2b3 - Monday July 20^th 1998

v2b2 - Sunday May 24^th 1998

v2b1 - Tuesday 10^th March 1998

v2a3 - Wednesday December 3^rd 1997

v2a2 - Friday October 10^th 1997

v2a1 - Friday August 15^th 1997

Version 1 timeline

v1.02 - Friday 6^th March 1998

v1.01 - Tuesday 23^rd December 1997

v1.00 - Thursday 10^th July 1997

2.6.2 Internet Drafts and RFCs

In addition to those drafts listed below, during the development of the CIPv3 implementation in ROADS version 2, feedback was provided to authors of a number of Internet Drafts in the FIND working group. The corrected drafts are on the IETF Internet Standards Track.

RFCs

2219 Use of DNS Aliases for Network Services. M. Hamilton, R. Wright. October 1997. (Format: TXT=17858 bytes) (Also BCP0017) (Status: BEST CURRENT PRACTICE)

Internet Drafts

WHOIS++ related (ASID working group)

WHOIS++ URL Specification, M. Hamilton, 03/13/1998, <draft-ietf-asid-whois-url-02.txt>

WHOIS++ templates, J. Knight, Patrik Faltstrom, M. Hamilton, Leslie Daigle, 03/16/1998, <draft-ietf-asid-whois-schema-03.txt>

Wide area service location (SVRLOC working group) Finding Stuff (How to discover services), M. Hamilton, P. Leach, R. Moats, 10/31/1997, <draft-ietf-svrloc-discovery-05.txt>

Advertising Services (Providing information to support service discovery), M. Hamilton, R. Moats, 10/15/1997, <draft-ietf-svrloc-advertising-03.txt>

Non-IETF working group Internet Drafts

Representing the Dublin Core within X.500, LDAP and CLDAP, J. Knight, M. Hamilton, R. Iannella, 07/28/1997, <draft-hamilton-dcxl-02.txt>

Cachebusting - cause and prevention, M. Hamilton, A. Daviel, 02/03/1998, <draft-hamilton-cachebusting-00.txt>

Distributing control of the Domain Name System, M. Hamilton, J Knight, 03/13/1998, <draft-hamilton-fix-dns-00.txt>

The Content-MD5-Origin: header, M. Hamilton, 03/13/1998, <draft-hamilton-content-md5-origin-00.txt>

2.5.5 Events organised

IRC virtual workshop for ROADS version 3 requirements gathering, Wednesday 6th and Thursday 7th of May, 1998

ROADS cataloguing guidelines meeting, Library and Learning Resource Centre, University of Bath, 17 September 1997

2.5.7 Conference presentations

Ferguson, N, SOSIG, ROADS and information gateways for Europe, Conference of German Professional Scientific Societies, Hamburg, March 1998

Ferguson, N and Russell, K, ROADS and possibilities for an International Mesh of information, CNI/NSF meeting, Minneapolis, October 1997

Ferguson, N, ROADS and the UK public libraries, EARL steering group, London, April 1998

Standards in the CHIC-Pilot Distributed Indexing Architecture, Peter Valkenburg, Dave Beckett, Martin Hamilton, Simon Wilkinson. (To appear) In proceedings TERENA Networking Conference, 1998.

Powell, A., Metadata. Internal staff presentation, Swiss National Library, Berne, 16 (?) September 1997, also: Library Science Talks, CERN, Geneva, Switzerland, September 1997.

Day, M., ROADS: metadata for resource discovery and interoperability. MetaData: qualifying Web objects, Universität Osnabrück, Germany, 13 October 1997.

Dempsey, L. and Powell, A., Metadata tutorial. Metadata Workshop, Luxembourg, 1 December 1997.

Heery, R., Position paper for EU-NSF Working Group on Metadata, Coalition for Networked Information, Washington DC, 2-3 February 1998.

Heery, R., The metadata landscape. Computing Laboratory Seminar, University of Kent at Canterbury, 24 February 1998.

Powell, A., Metadata for the Web. UCISA-TLIG: New Opportunities - Information Services for the Next Millennium, University of Southampton, 31 March 1998.

Day, M., Metadata developments in the library community. Digital workflow through production and documentation: a seminar organised by the Documentation Commission of FIAT/IFTA, BBC Conference Centre, London, 7 May 1998.

Day, M., UKOLN Metadata Projects. META-LIB-Workshop, Niedersächsischen Staats- und Universitätsbibliothek (SUB), Göttingen, Germany, 23 June 1998.

Powell, A., Metadata for the Web: RDF and the Dublin Core. UKOLUG 20th Conference, Manchester Conference Centre, 15 July 1998.

2.5.8 Publications

Hiom, DA, "Roads to Metadata", Proceedings of IASSIST '97, 1997

Dempsey, L, Heery, R, Hamilton, M, Hiom, DA, Knight, JP, Koch, T, Peerboom, M and Powell, A, A review of Metadata: survey of current resource description formats

<URL: http://www.ukoln.ac.uk/metadata/desire/overview/ >

Knight, J.P., 'HTML - Which Version?', in The Cybrarian's Manual, Ensor, P. (ed), American Library Association, Chicago and London , 1997, pp 206-208, ISBN 0-8389-0693-1.

Kirriemuir, J., Brickley, D., Welsh, S., Knight, J.P. and Hamilton, M.T., "Cross-Searching Subject Gateways: The Query Routing and Forward Knowledge [Approach"], D-Lib Magazine , January 1998, [online], ISSN 1082-9873.

Dempsey, L. and Heery, R., "Metadata: a current view of practice and issues". Journal of Documentation, Vol. 54, no.2, March 1998, pp. 145-172.

Dempsey, L., Russell, R. and Heery, R., "In at the shallow end: metadata and cross-domain resource discovery". In: Discovering online resources across the humanities: a practical implementation of the Dublin Core, edited by Paul Miller and Daniel Greenstein. Bath: UKOLN on behalf of the Arts and Humanities Data Service, October 1997, pp. 63-71. <URL: http://ahds.ac.uk/public/metadata/disc_07.html >

Heery, R., "Naming names: metadata registries". Ariadne (Web version), No. 11, September 1997. <URL: http://www.ariadne.ac.uk/issue11/metadata/ >

Heery, R., "What Is ... RDF?" Ariadne (Web version), No. 14, March 1998. <URL: http://www.ariadne.ac.uk/issue14/what-is/ >

Heery, R., Powell, A. and Day, M., Metadata. Library & Information Briefings, No. 75. London: South Bank University, Library Information Technology Centre, September 1997.

Heery, R., Powell, A. and Day, M., "CrossROADS and interoperability". Ariadne (Web version), No. 14, March 1998. <URL: http://www.ariadne.ac.uk/issue14/metadata/ >

Powell, A., "An idiot's guide to Dublin Core". New Review of Information Networking, Vol. 3, 1997, pp. 157-164.

2.7 Particular successes

2.7.1 The modularization of ROADS and the growing ROADS community

ROADS seems to be becoming increasingly popular outside of eLib funded ANR services. This is to the benefit of ANR users, as we are rapidly approaching the point where the sheer number of existing ROADS-based services acts as a mechanism to increase use and take-up. Also, the larger the ROADS user community, the greater the pool of expertise which can be exploited by the ANR services.

The tool-kit approach of allowing people to select which of the ROADS tools they wish to use has contributed to this success.

It is hoped that the cleaner ROADS API which will be delivered in version 3 will allow these people to further enhance and customise ROADS tools and then feed enhancements back into the development process for use among the wider community.

2.7.2 Assisted take-up

One of the major successes of the ROADS project has been in the area of `assisted take-up',.

The development of the ROADS user community focused around the various mailing lists has greatly facilitated this aspect of the service. ROADS has fostered this approach from the very beginning however, and we are now in the position where we have a vibrant user community, who assist each other and actively participate in the encouragement of further take-up of ROADS software.

In the next year we will extend this approach with the modularization of our code base, and will also be developing a range of `How-To' guides offering suggested approaches to implementing gateways using the ROADS tool-kit. We will also be arranging various `gatherings' of the user community to foster the sharing of expertise.

2.7.3 Internationalisation of the project

One of the particular successes of ROADS this year has been the internationalisation of the project, with ROADS playing a leading role in bringing together many, globally remote, resource discovery services. During the API virtual workshop for example, we had participants from the US and mainland Europe. On ROADS mailing lists, there have been queries and responses from people in the US, Australia, Finland, Holland, Denmark and several other European countries.

Whilst ROADS is funded to provide a service for the UK community, this international collaboration and take-up has been enormously beneficial because it has made the work of the ROADS project more widely known in the global resource discovery community. This has in turn led to greater dissemination of the work of the services that use ROADS software, such as the eLib ANR services, the project funders, and programme executive (JISC and eLib).

This collaboration has complemented several other international initiatives, most notably the DESIRE project mentioned elsewhere in this report.

The ROADS project has therefore proven useful as an international dissemination vehicle for the resource discovery components of the eLib programme and other JISC-funded initiatives. It was notable that several of the overseas delegates at the recent UKOLN conference on information landscapes first became aware of the resource discovery activities in the UK through ROADS, and ROADS publicity of the services which it supports.

This broadening of the scope of the ROADS community has also meant that the project has had to adapt to fit in to the wider context of developing international standards. For example, the work of Dan Brickley on the development of the Resource Description Framework (RDF) with the World Wide Web Consortium has meant re-evaluating core ROADS strategies. This means that ROADS is now better placed to offer guidance and support to ANR services to ensure they can respond to emerging standards and maintain their position at the forefront of developments in networked resource discovery and access provision.

3. Learning from the process of implementation

3.1 Difficulties

3.1.1 Changes in personnel

There have been several changes in personnel at Bristol over the past year. This began with the departure of Paul Hoffman (ROADS project manager) in March 1998. This resulted in a restructuring of the project staff at Bristol, and the ROADS liaison and co-ordination effort was spread more widely among all the members of the Bristol team with John Kirriemuir taking on overall responsibility. This re-organisation paid dividends in that the management of the project was de-centralised and involved project staff taking greater responsibility for key areas of effort. This in turn led to greater co-ordination and better communication between the three project partners.

John has recently left to become project manager of the OMNI ANR service. This means that although he is no longer working directly for the project, he is still a very active member of the ROADS community and can therefore give the ROADS team the benefit of his experiences running a ROADS service.

John's role was taken over by Paul Hollands at the beginning of August 1998. Paul is a member of ILRT staff who previously worked for SOSIG both as a trainer and a cataloguer. Paul also worked at Loughborough University prior to joining the SOSIG team and knows the ROADS team there.

Paul hopes to use his experience as a SOSIG cataloguer, and knowledge of gateway user needs gathered during training, to further increase the user focus of ROADS software development and liaison work.

3.1.2 Working with geographically disparate partners

On several occasions over the past year, it has been difficult to arrange meetings involving all the ROADS partners. This is partly due to the growth in the general ROADS community but is also an inevitable effect of having a geographically disparate project group.

To alleviate this situation we have been investigating the use of various networked technologies, most notably the various ROADS mailing lists and IRC, and these have not only provided neat solutions but also revealed unexpected benefits, which are outlined in section 2.1.2.

3.1.3 Growing support demands on the project staff

The increased take-up of ROADS by non-ANR services has brought many benefits to the ANR projects both directly and indirectly. With these benefits has also come an increase in the support burden on ROADS, particular for the Loughborough team.

Although this has not yet reached a stage where it is reducing our capacity to fulfil our stated objectives, it is clear that the ROADS support strategy needs to be clarified before this becomes a problem.

Priority must always be given to the ANR services we are funded to serve. However, these services would benefit from a more clearly developed system for ROADS support as well.

The issue of support (with clear priority being given to ANR services) will be addressed in the near future.

3.2 The influence of other projects

3.2.1 The SOSIG and Biz/ed gateways

Because members of the Bristol ROADS team also work on several other ANR services, particular, SOSIG and Biz/ed this means that we are ideally placed to `bench test' any interesting new ROADS developments quickly and then feed that back into the development cycle. This means that ROADS has a strong user focus and that development work is done with a clear understanding of subject gateway requirements.

These `synergies' are also exploited in other ways. For example a key mechanism for dissemination of information about ROADS is the publicity and training effort of these subject gateways. This also means that we have access to direct feedback from gateway users through training etc.

The ROADS team also have access to live services on which to test new developments. This has proven invaluable.

3.2.2 DESIRE

Becoming part of the DESIRE programme has created an increased demand for ROADS technologies and prompted new developments such as the multi-lingual functionality in ROADS version 2 and new resource description guidelines and standards.

3.2.3 International standards bodies

Members of the ROADS team, both through ROADS funded work and other projects that they work on, are active in various standards making bodies, such as IETF, the W3C and the Dublin Core community.

The work of all three partner groups on the Dublin Core initiative has meant that ROADS has had a broad level of involvement in the development of this important international standard and this has led, in turn, to the development of new ROADS tools, as well as ensuring ROADS remains compliant with emerging standards.

Perhaps even more important in the long term is the fact that ROADS staff are members of various W3C working groups. This has led to collaboration with resource discovery experts in the US, Europe and Australia as well as co-operation with leading corporate developers such as Netscape.

3.3 Changes to the plan in the light of experience

3.4 Unexpected results and opportunities

John Kirriemuir did several presentations in Helsinki during February/March 1998 with Kelly Russell of the eLib office. These were aimed at disseminating information about eLib, ROADS and the ANR services to the Finnish Library and Information Science Communities.

The presentations were extremely well attended, some containing over a hundred strategically important people, such as University librarians, computer centre directors, and library school directors.

Two presentations were given exclusively on ROADS, reaching a combined audience of over 80% of Finnish University library directors. Partially as a consequence of these presentations, and partially as a result of previous liaison by the ROADS project and, in particular, by Emma Worsfold of the SOSIG ANR service, Finland has decided to adopt ROADS as the `Resource Discovery Gateway System of choice'.

3.5 Changes in understanding and perceptions

3.5.1 Continuing use of the WHOIS++ protocol for ROADS version 3

WHOIS++ was chosen originally because of its simplicity and because it was a proposed Internet standard which provided powerful searching facility for distributed databases using forward-knowledge based query routing and centroid technology.

WHOIS++ is a mature technology which means that it is also robust and easy to maintain. The ROADS team recognise however that new standards are constantly arising, developments such as LDAP, RDF, CORBA, Dublin Core and so forth. It is also clear that profile of the Z39.50 protocol continues to rise.

Because of the robustness and maturity of WHOIS++ ROADS has retained use of it as the basis of our tool set into version 3. We have however, also responded to the growth in influence of other protocols as demonstrated by our success in moving ever closer to full interoperability with Z39.50.

This work has revealed that Z39.50 presents a number of problems if it is to be applied to the types of environments and tasks that the ROADS software seeks to address. One very grave drawback to using Z39.50 is that each implementation of the protocol in commercially produced systems seems to differ from the next. Z39.50 also lacks the built-in capability to generate and process any kind of forward-knowledge, which makes it unsuitable for query routing over large distributed databases without additional supporting servers.

Further information about the problems raised by the use of Z39.50 in a distributed search environment can be found in the following article:

Rusbridge, C "Towards the Hybrid Library", D-Lib Magazine, July 1998

<URL: http://www.dlib.org/dlib/july98/rusbridge/07rusbridge.html>

We have therefore chosen to opt for an approach which retains the benefits of WHOIS++ over Z39.50 but also allows for a level of interoperability between the two protocols.

With members of our team participating in the W3C we are keeping a watchful eye on developments particular in the RDF and XML standards and there has been much discussion about moving ROADS to these technologies as and when it is felt the standards are mature enough.

As far as ROADS developers are concerned WHOIS++ was chosen as a technology to use until something better came along. However, developments that we have made over the past year have proven the suitability of the protocol for fulfilling the need of the ANR services for version 3 and beyond.

3.5.2 Shift in approaches to development

Over the past year the ROADS team have also changed their approach to the way that the various tools and systems are developed and bundled. This is to allow for the emergence of greater user participation in software development and bug fixing. With version 3 and the ROADS API we are moving towards a more modular approach to the software, which would, for instance, allow service providers to set up ROADS gateways using a variety of different types of back-end databases while still providing the core of the useful ROADS interoperability features.

The actual structure of the software has also changed over the past year and now conforms much more to the emerging object-oriented features of Perl 5.

Improvements in the ease of installation and configuration of ROADS as well as the many improvements in areas such as the cataloguing interface have reflected a greater user focus in our development strategy. ROADS 2 has been built to be much more user friendly than previous versions and this process will continue in version 3 and beyond.

A side effect of this has been that it is now much easier to tailor ROADS for purposes other than the standard subject gateway for which the software was developed.

4. Evaluation results

4.1 ROADS Evaluation in 1998/99

An informal evaluation of the training and documentation needs of new, and potential, ROADS users was made in April-July 1998. This sought to discover any gaps in the documentation (content and format), training, Web site, mailing lists and other forms of ROADS discussion media. Many useful suggestions came out of this evaluation, and these are currently being discussed by the ROADS team.

The most popular idea was for a workshop for those who wish to establish new ROADS-based gateways. This would concentrate on the basics , i.e. non-technical explanations of the inner workings of ROADS, how to set up a gateway on a variety of platforms, simple configuration and customisation and so forth.

This workshop is an ideal opportunity for ROADS to reach a wider audience, and directly increase the number of services committed to producing high quality resource descriptions within the UK academic and research communities (an action which can indirectly benefit such initiatives as the DESIRE II project, the JISC-funded DNER etc.)

The results of this informal evaluation are attached as Appendix A.

ROADS will also undergo a formal evaluation in the coming year. This is being facilitated by the IMPEL2 project. The format of the evaluation is still under discussion; however, the main emphasis is to see if ROADS:

The process of evaluating ROADS will involve interviewing various types of ROADS `users', including:

ROADS evaluation by IMPEL2 is currently due to start in January 1999 and occupy the evaluator for a half to one day per week. A report is expected in the late spring/early summer, subject to negotiation.

5. Future developments

5.1 Future funding and development structures

There has been considerable uncertainty over funding for the ANR services over this reporting period and this has had a direct bearing on the future of the ROADS project.

JISC has held back any decision on long term funding of ANR services until the various models for a possible Resource Discovery Network are clarified.

The writing of this report also coincides with the final draft of the Haynes report on the future of the ANR services, and so any discussion of the future direction and funding of the ROADS project, and hence the development of the ROADS tool-kit, has to remain tentative at best and focus on short term objectives.

It has become clear that the commercial sale of the ROADS tool-kit would go against the open development approach which the project has fostered so successfully. The main strengths of the current ROADS tool-kit are that it is highly focused on the needs of the ANR subject gateways and that it is based on an open architecture. Future development is dependent on it being free, its source code being available, its compliance to open standards, and its having a committed, distributed team of developers. The benefits of this approach have been demonstrated by the success of the Linux operating system which was the first set of software tools to be developed in such a way.

Unfortunately, these very strengths predicate against the ROADS tool-kit being a success as a commercial piece of software.

One possible alternative stream of revenue may be from selling ROADS related advice and consultancy. This would not provide enough money to maintain the project staffing effort at a suitable level to support the ANR services as we do currently however.

5.2 Supporting ANR services in the future

Over the next year the ROADS will continue the tasks we have already begun such as the release and implementation of version 2 and the development of version 3. The funding for the next year is interim and for maintenance of the project until the future of the ANR services is decided. Bearing in mind new development work will be constrained somewhat by these lower levels of funding, we will be investing possible effort in the following areas over the next year:

5.2.1 Survey to improve the user feedback process

The ROADS liaison team will be making a survey of current users (where possible making visits) as part of a review of the successfulness of ROADS 2 and the functional requirements for ROADS version 3. More formal feedback and support mechanisms will be implemented in the light of these findings.

5.2.2 Addressing ROADS support

The ROADS team will begin a process of consultation with the wider user community about the support requirements of fledgling ROADS services. Various options will be explored including possible third party support for non-ANR ROADS users.

5.2.3 Contacts database of ROADS users

The ROADS liaison team will be setting up a contacts database of current and potential ROADS users. This will be done to improve the tracking of queries and provide faster responses to problems arising.

5.2.4 Improving administrative functions

Much of the effort this year has been focused on improving the cataloguing user interface and giving cataloguers more options. This effort will be extended to improving other aspects of gateway administration such as link-checking and the process of handling and responding to new resource submissions. It is envisaged that this process will begin with the survey outlined above. It is not clear whether this would result in a set of software tools or simply a set of recommendations. This depends on the level of requirement from ANR services and others.

5.2.5 Management information and reporting

One aspect of the ROADS tool-kit which has not been developed but which might prove useful is a more systematised approach to extracting management information and statistics from ROADS gateways. This needs to be investigated, although it is not clear whether this would result in a set of software tools or simply a set of recommendations. This depends on the level of requirement from ANR services and others.

It may be possible to use third party tools in a solution.

5.2.6 ROADS and DESIRE II

DESIRE II promises to feed into the ROADS project and offer many benefits. DESIRE II is being led by the ILRT with the participation of UKOLN and other European partners.

5.2.7 Scaleability testing

ROADS needs to be tested for scaleability with datasets in the order of tens of thousands of records. In the near future a ROADS database of 60,000 complex bibliographic records will be set up and tested at Bristol. This will build on the work already done with the Harvest demonstrator of 40,000 records currently being maintained by the ILRT.

ROADS appears scaleable in terms of searching at these orders of magnitude. Cataloguing and re-indexing are a problem however. Solutions such as distributing resources over several ROADS servers using cross-searching will be investigated.

The impact of different hardware specifications will also be tested.

5.3 Interoperability

5.3.1 Interoperability with Windows NT systems

ROADS is a UNIX only set of tools. However we recognise that some users may wish their ROADS services to interact with Windows NT based systems or wish to add some degree of ROADS functionality to them.

Work is in progress, investigating several ways this might happen. Techniques are being developed to allow users to catalogue resources using Microsoft Access, either via a direct connection to a ROADS database via ODBC, or by making periodic exports in comma separated text files, which are then used to create ROADS templates automatically.

Various cross-searching approaches are also being tried, via ODBC and also by `tunnelling' WHOIS++ queries over HTTP and supplying responses using the Active Server Page technology available with the Microsoft IIS Web server.

All of these possible developments have been made easier by the new ROADS API.

5.3.2 Cross-browsing

Work is in progress with the Biz/ed subject gateway which is investigating techniques of `cross-browsing' distributed resource collections.

5.3.3 RDF and XML

ROADS is already interoperable with a number of other systems and this trend will continue. The `next big thing' is the rapid development and acceptance of XML, RDF and Dublin Core by the W3C member companies. If these technologies are accepted rapidly, then ROADS and thus, the ANR services, will be well positioned to make use of them. For example, with Harvesting technology already being tested with ROADS version 2, an XML document containing Dublin Core metadata in RDF could be retrieved and processed to extract that metadata and speed cataloguing. Conversely, a ROADS database could hold metadata which could then be presented as RDF and served over the Web. (Netscape has already developed utilities to render such information in a meaningful, human-readable way.)

RDF and XML will need to be supported in the near future. We will also be watching with interest the development of RDF based search and retrieval protocols and CORBA.

Members of the ROADS team are active participants in the development of the RDF standard through work being done with the W3C. There is work in progress in collaboration with the Biz/ed service to provide RDF based searching and browsing tools.

A demonstration of this work is visible by using the RDF site map facilities in Netscape navigator 5 on the following URL:

<URL: http://localhost/discovery/demos/rdf/ >

Dan Brickley of ILRT, through Bristol's membership of the World Wide Web Consortium (W3C), has been heavily involved with the development of the Schema specification language. Dan is co-editor of the RDF Schemas specification document, with Andrew Layman of Microsoft and R.V.Guha of Apple [cite]. It is expected that the document will be completed by early autumn, at which point the focus of this work will likely change to address technologies for searching and browsing RDF repositories.

5.3.4 RDF, Warwick Framework and Reggie

RDF Schemas represent an evolution of the Warwick Framework modular approach towards metadata vocabulary management. Schemas provide machine readable definitions for metadata vocabularies, and can be used to automate tools for data entry, indexing and validation.

To illustrate the value of a schemas-based approach to metadata definition, the ILRT metadata team have recently been experimenting with the use of the Reggie metadata creation applet in a ROADS context (Reggie has been developed by the Resource Discovery Unit at the Distributed Systems Technology Centre in Australia). By defining a formal schema for the ROADS template types, Reggie can be used 'out of the box' to support the creation and editing of ROADS records, exporting in the RDF format. This work has only begun recently (Aug 1998) but appears promising. The creation of a full RDF Schema for ROADS record types can not clearly proceed until the Dublin Core data model work has been completed.

A number of ROADS team members are contributing to the DC data model effort; when this is complete, we should be able to use RDF Schemas to create a ROADS vocabulary which elaborates on the basic DC in the context of our particular needs. RDF data entry tools such as Reggie or Netscape's Aurora system in Mozilla will then be useable tools for use in an Internet cataloguing context. This is particularly interesting in the Netscape case since the source-code to their RDF database and user-interface is freely available. One exciting development is the use by Netscape of LDAP servers for storing individual's RDF-based bookmark databases. Since Netscape's bookmark system is now using RDF, this raises the prospect of thousands of individuals having a sophisticated metadata management tool as part of their normal information environment. The implications of this for the Subject Gateways are not yet entirely clear: one model would be to build upon the 'trusted information provider' approach, allowing distributed groups of subject experts to catalogue into a central database via their familiar bookmark management interface.

The ROADS liaison team are following various subject-gateway experiments in this area with interest, advising and collaborating where appropriate. As in the case of Harvesting, it is often the case that ROADS itself lacks the resources to explore all possible approaches: the subject-gateway model and its environment are evolving at an alarming speed. It is consequently important that ROADS users are provided with fora such as roads-hackers and the ROADS newsletter in which to share their explorations of these new areas. Example of this include the hierarchical subject-listings browser developed by SOSIG and the RDF site map and `Yahoo-like' interfaces built by Biz/ed.

Various activities under phase 2 of the DESIRE project will explore the application of RDF to distributed indexing and cataloguing scenarios. The ROADS team liaise very closely with the DESIRE project partners, and will be tracking the progress of the DESIRE RDF work. Other ILRT projects such as the eLib funded Biz/ed gateway have been exploring RDF based solutions to cross-browsing and collections-level description issues. These will be disseminated to other ROADS projects through the ROADS liaison activity.

RDF is increasingly of interest to the wider computing community, particularly since it is backed by both Netscape and Microsoft. We expect to produce, in collaboration with the subject gateways, some simple lightweight demonstrations -- eg of RDF Sitemaps for browsing the subject listings -- which could be used to raise the public profile of the subject gateway movement and to evaluate the advisability of future more substantial work in this area. The ILRT-managed RDF Developers list, rdf-dev, for example, has a ROADS-based resources database in preparation. This will be used as a showcase installation and to pilot any RDF-related add-ons to ROADS contributed by other gateways.

5.3.5 Evaluation of centroid technology

Analysis will need to be made of the usefulness of centroids to ANR services for forward-knowledge query routing. This information can be used to formulate new CIP v3 index types which may then be used not only for subject gateways but also for the vast robot harvested databases currently prevalent on the Web.

6. Appendix A - Informal study of ROADS Training and Documentation

John Kirriemuir - April 1998

6.1.1 Version

This study supersedes the previous study of the summer of 1997 that was carried out on ROADS training and documentation support.

6.1.2 Purpose of this study

The aim of this study was to discover potential gaps in the training and documentation support that the ROADS project provides for its core user community. Asking users who were already familiar with the project, its associated software, training and documentation support would be counter productive. We needed users who were looking at ROADS for the first time. Consequently, the study is restricted to those users who are new to establishing ROADS-based systems.

6.1.3 Methods and targets of study

Those who contributed to the study included:

Comments were solicited informally and off the record. Initial comments were located from:

Interesting comments were then followed up.

It should be noted that some of the users who provided comments and suggestions were not from the UK community. Whilst their documentation needs may be similar to like-minded developers in the UK, it is unlikely that overseas people could participate in face-to-face workshops and the like.

6.2 The guide to ROADS, help guides and FAQ's

Several users pointed out that the guide to ROADS, available from the documentation section of the ROADS Web site, though useful, was starting to look somewhat dated. The guide needs to updated more regularly. In addition, it needs to be expanded to cover areas such as:

Several people commented that the ROADS guide was pitched at the right level of technical expertise for users who were looking at using the software.

It was suggested by several people that separate, concise, guides for specific ROADS-related tasks should be produced. These would be more technical than the ROADS guide, and would therefore be more practical and "hands-on" in nature. Topics that were suggested were:

It was also suggested that ROADS should produce a "protocol position statement", which would outline what the current and future positions of the ROADS project in relation to accessing gateways via various protocols e.g. Z39.50, WHOIS++, LDAP and so forth.

The FAQ guide to ROADS was only mentioned by one person, and seems to be something which most people overlook when they visit the ROADS site. Some users were unaware of its existence. When asked, they stated that an FAQ would be a good thing, though one person pointed out that it would difficult to decide who it was aimed at, as some people looking to use ROADS are quite technically minded, whereas others are managers, end-user support people, or librarians.

6.3 Training "workshops"

The most repeated suggestion by users was to hold a workshop for people relatively new to ROADS. Indeed, this suggestion was iterated publicly by users from the Open University and Leeds University ROADS-based gateway development teams on the open-roads mailing list.

The workshop would be aimed at people who were either just starting to use ROADS, or who were thinking seriously of using it. The audience would, therefore, not contain people who were overly familiar with either the ROADS software or documentation. Two users pointed out that a workshop audience containing ANR people would be counterproductive, as it may inhibit people less familiar with ROADS from asking "basic" questions, and the verbal sessions may be "hijacked" with discussions or questions beyond the scope of the workshop. Users felt that the workshop would be more effective if it was clearly aimed at people less familiar with ROADS. One person suggested that a warm-up presentation at the start of the session by an "experienced ROADS user" from an ANR service would be useful.

Several users stated that they would prefer the workshop to be led by either Jon or Martin (the Loughborough team and main software developers). This was because "they were very helpful", "were great at explaining technical things in a relatively non-technical way" and were not "threatening" or "knowledge snobs" (sic.).

The timing of the workshop was mentioned by a few users. Many of our target audience felt Autmun 1998 would be ideal in terms of their development time scales.

The actual content of the workshop should be determined mainly by the attendees. Topics that have been suggested include:

The workshop should be practical in nature, with the various developers having access to ROADS on a computer in front of them. A hands-on workshop would also have the advantage that people could alternate between a demo of ROADS and, via telnet, their own installation.

The level of technical knowledge of the attendees of such a workshop was also mentioned by two of the people who would be keen to attend. It was accepted that attendees should have at least a grasp of basic UNIX, without which it would be difficult to understand many of the concepts of how ROADS works. However, it was firmly expressed that it should not be a precondition that a knowledge of Perl was required, as many of the users would be put off attending, and there was no need for the workshop to "sink to the level of the software code" (sic).

From the point of view of ROADS, such a workshop would be a good idea. for several reasons:

6.4 Advanced workshops

Several users suggested that they would like a more advanced workshop at some point in the future, perhaps six months after the beginners workshop. This would cover topics such as:

Such a workshop would have the advantage of attracting the more technically aware users who are currently either setting up, or thinking seriously about setting up, a ROADS-based gateway, as well as the core ANR services.

6.5 IRC sessions

It was mooted by several people who had either heard about, or participated in, the API IRC session, that such a mechanism is a good tool for meetings of disparate groups Two people pointed out the attractiveness of having an automatically produced log of the session.

The greatest benefit of IRC is that it is the only practical real-time medium via which non-UK people can participate (good quality videoconferencing is too tricky). The API IRC session involved participants from the US and mainland Europe, and was very appreciated by them. With the number of overseas ROADS developers increasing at a steady rate, IRC should perhaps become a standard tool for discussions between ROADS users (as one user in Finland pointed out, there is no better practical alternative)

Suggestions for future IRC sessions included:

In addition, the use of IRC is a cost-effective form of communication and can add indirectly to a projects dissemination effort, both factors which the projects funders have encouraged in the eLib programme in the past.

6.6 Rationalisation of the ROADS Web site

There were a variety of minor comments about the ROADS Web site. Most people liked the design, and especially the consistent navigation between the three sites. When questioned, two people were surprised to discover that the ROADS Web pages are not, as they had thought, kept on one Web site but spread amongst three.

The comments about the ROADS Web site included:

6.7 Mailing list support for new ROADS users

The various users who expressed an opinion were very positive about the set-up regarding mailing lists for ROADS support. Frequent mention was made of the helpfulness of especially Jon and Martin in providing speedy, comprehensive, friendly and readable replies to queries.

A few minor points about ROADS mailing lists were made:

6.8 Recommendations for future surveys

It is recommended that the training and documentation needs of potential, or new, ROADS users are evaluated periodically by the ROADS project. As ROADS is going to be evaluated by the IMPEL2 eLib project during the first half of 1999, such evaluation can form part of that study.

A future study could also possibly encompass users who have been interested in using ROADS, but who have not made themselves known through email enquiries, as well as those people who have downloaded the ROADS software but have not sent in follow-up enquiries to the ROADS team. Getting responses from these users is tricky; web-based questionnaires and similar mechanisms could be employed, but they may not prove popular. Another more contentious mechanism for identifying specific users would be to make access to the ROADS software dependent on the submission of details of the interested party, via either a Web-based fill-in form system, or through a direct "contact us for the software" approach (both of these mechanisms would affect the image of ROADS and the amount of take-up of the software).

6.9 Additional comments

The suggestions discussed in this report would, if they were all implemented, take a significant amount of resources. In addition, the workshops might also incur costs for room and PC hire etc. It would be prudent for the ROADS project, in its forthcoming internal review between the project partners, to take a close look at what resources it has and whether they need to be reallocated.

7. Appendix B - Financial Report(To follow)