1. Overview
1.1 The Problem
1.2 Background: HILT Phase I to HILT Phase II
1.3 Staying in Step with other Relevant JISC Activities
1.4 An Experienced Team and a Rich Distributed Research Environment
2. Project Aims
3. Project Description, Including General Approach and Methodology
3.1 General Points

Appendix A: Interactive Terminologies Route Map (TeRM) Diagram
Appendix B: Key Questions Relating to Terminology Map Modelling Requirements
Appendix C: Relationship to HILT Phase 1, the DNER, and the JISC Information Environment

HILT Phase I reported in November 2001. It was funded jointly by JISC and RSLP and was in line with the 'Study and Specification' stage of one aspect of the Shared Services programme - that aspect relating to the provision of thesauri and terminology services. HILT Phase II moves this process into the 'Pilot Project' stage, focussing - as recommended by the HILT Phase I evaluator - on terminology and thesauri requirements at the collection level, but also bearing in mind the need to extend this in due course to the needs of item level retrieval. Phase II will last for 12 months, will cost £89,000, and will utilise the work of HILT Phase I, and the skills and experience of the team that carried it out, to:

• Build an initial pilot, then develop it in line with user, service, and expert evaluator outputs
• Determine detailed requirements, costs, and benefits to FE and HE users of a full terminologies service focussed primarily on collection level needs

HILT Phase II will set up a pilot terminologies route map or TeRM service, similar to that proposed in HILT Phase I (see Appendix A for description), aiming to:

a. Provide a practical experimental focus within which to investigate and establish subject terminology service requirements for the JISC Information Environment, with particular reference to DNER, RDN, User, Collection Level, International Compatibility, and local, regional, national and UK-wide access considerations.

b. Make recommendations as regards a possible future service, taking into account a range of factors, including the level and nature of user need, practicality, design requirements, effectiveness, functionality available in existing commercial software packages as against original development, and (above all) costs against benefits.

There are three reasons for using a specific piece of commercial software at this stage:

1. Experience within HILT PHASE I suggests that project participants find it easier to discuss the requirements of such a service given a real illustrative example on which to focus. It is therefore believed essential that we mount an illustrative pilot early on in the project in order to help engage the interest and attention of users and other stakeholders and give them a practical environment within which to envisage and consider the problem. Wordmap is being used because we want to have a real working demonstrator at an early stage for users and service providers to interact with. Attempting to draw out the full requirement before implementing an illustrative pilot would, it is believed, result in a poorly researched requirement as users and service providers would not have been sufficiently stimulated by operation in a real context to allow a full specification to emerge.

2. It is feasible to negotiate a low price for a pilot of this kind with one company before the project, it would not be possible to guarantee doing this in the middle of a project after we'd spent many months preparing a requirement.

3. Taking a development from scratch approach at this stage would almost certainly involve employing an experienced programmer for the duration of the project. This would be in addition to the one full-time member of staff costed below and would certainly cost more than the cost of the software. It would also entail additional risks. On the one hand, such staff are hard to find and often move on for improved pay and conditions before a project is completed. On the other, hold-ups or failures in software development schedules could undermine the schedule in respect of other project activities. The Wordmap approach is a pragmatic one that will enable us to evaluate the real uses and issues in a timely way, whilst avoiding the potential waste and risk involved in development from scratch before a full requirement has been established.

The initial illustrative TeRM would be based on the RDN terminologies , on terminologies available as part of the Wordmap taxonomies set, which include, in particular, a set of terms used by general internet users, and on selective subsets of LCSH, DDC, UNESCO, and AAT. OCLC will provide an LCSH - DDC mapping, and may also be able to provide a DDC to Conspectus subject headings mapping. The UNESCO thesaurus is available online and we will look to obtain AAT selections from manual sources. The aim would be a selective mapping sufficient for the purposes of the pilot in the first instance - i.e. not a comprehensive terminologies map. Consideration would also be given to the points noted in Appendix B below and to the various issues raised by the HILT Phase I evaluator, Leonard Will (HILT Final Report , Section 10).

1. Set up: Web-site activities begun; Service, User, Community stakeholders e-lists set-up.

2. Refine and begin to implement project plan: Begin professional level evaluation; Fully scope problem; Identify evaluation criteria for success or failure of the TeRM approach, looking at technological, terminological, cost, and user requirements; Determine subject areas to focus on; Begin to identify commercial software suppliers; Involve evaluator and the second terminology expert fully in all processes.

3. 'Engage' with users, service staff, communities, and the problem in general; implement illustrative stage 1 pilot: Start dissemination activities; set focus group dates; Install pilot software; Implement illustrative (stage 1) pilot to inform and stimulate discussion; Plan focus groups with a view to drawing out initial service specification; Examine the process of assigning subject terms at RDN hubs and at other relevant sites; Begin general and M2M requirements studies; Begin functionality of commercial software survey; Hold focus groups (include terminology experts).

4. Begin to build stage 2 pilot TeRM based on illustrative pilot: Determine initial service specification via focus groups, professional level evaluation, general and M2M requirements study, functionality survey, and project team analysis and discussion; Implement operational (stage 2) Pilot Service.

5. Terminology map modelling for TeRM (starts prior to implementation of stage 1 pilot): Investigate terminological and functionality requirements of a TeRM through intellectual analysis, the general and M2M requirements studies, and empirical testing in a practical environment. Utilise the following process when conducting empirical testing

a. New user authenticates as either HE or FE then chooses from the range of available entry points: local, regional, national, UK-wide (general) and UK-wide (subject hub).
b. New user 'disambiguates' her subject requirements using interaction with TeRM
c. Appropriate terms are sent to the HILT Phase II CLD-based landscaping mechanism
d. HILT Phase II landscaping mechanism builds appropriate entry landscapes, showing only those collections appropriate to the entry point in question. In some cases, these will be 'mock-ups' of the interfaces of existing services or subsets of those services:

• An FE college or university OPAC screen with a small change to permit the further discovery of collections held elsewhere
• An RDN subject hub (e.g. EEVL) or specific subject page (e.g. engineering in EEVL)
• RDN 'Central'

e. In other cases, they will be CAIRNS-type geographical or subject landscapes:

• The 3 Glasgow Universities
• A computer-aided design subject strength landscape showing Strathclyde University and Heriot Watt University as shown by accessing this URL: http://cairns.lib.strath.ac.uk;cfdocs;CAIRNS;CAIRNS;SubjCollSelect.cfm?uSubID=3128
• A 'mock-up' of an DNER-wide level entry page, dynamically generated from the SCONE copy database which would have a selection of additional DNER collections added for the purpose

f. TeRM terminologies route map is adjusted until the CLD-based landscaping operates as it should at this relatively simple level
g. A representative set of single term subject queries are then 'disambiguated' and 'mapped' via the TeRM and then run against the collections in each of these base landscapes, the latter process utilising CAIRNS cross-searching where the collections have Z39.50 catalogues or service by service methods where cross-searching is not currently possible (e.g. for the local OAI database). h. TeRM terminologies route map is adjusted until the single query item-level searching operates as it should.
i. Each base landscape has an 'expand collections searched' button. When this is clicked, a new TeRM-mediated landscape is generated depending on the nature of the subject query being processed, the user (HE or FE), and the base landscape (e.g. institution-level would go to metropolitan area level, RDN subject hub to DNER level, and so on). Again - as in e above - the TeRM terminologies route map is adjusted until the CLD-based landscaping operates as it should at this second collection discovery level.
j. The same representative set of single term subject queries utilised at h above are again 'disambiguated' and 'mapped' via the TeRM and then run against the collections in each of these new 'second-level' collections landscapes, the latter process once again utilising CAIRNS cross-searching where the collections have Z39.50 catalogues or service by service methods where cross-searching is not currently possible. The TeRM terminologies route map is again adjusted until the single query item-level searching operates as it should for item-level cross-searching at this second level.

Notes 1. There are two key questions underlying the above investigation of terminology map modelling requirements - see Appendix B for details
2. The above process will be monitored and influenced by the ongoing professional level evaluation process

6. Gradually optimise effectiveness and functionality of stage 2 pilot to produce stage 3 pilot and interim specification for a future service, utilising this same process

7. Investigate costs of an operational service at various service levels: Determine interim specification as described above; determine associated terminology-related set-up and maintenance costs; estimate feasibility and cost of a full service based on commercial software or adapted commercial software; estimate and compare likely cost of HE/FE developed software or any available open source software.

8. Prepare, disseminate, and obtain feedback on, Interim Report

9. Complete, launch, and conduct user trials on, improved stage 3 pilot: finalise design based on interim report feedback and ongoing professional level evaluation; Launch improved pilot; Plan user trials and focus groups; Hold user trials and focus groups

10. Finalise user, terminological, and functional requirements of a future service and refine estimated set-up and maintenance costs; further improve pilot if necessary and possible.

11. Project completion activities: Compile draft final report; Evaluate project and pilot and encompass results in draft report; Circulate draft final report and obtain final feedback; Finalise report; Submit and disseminate Report; Close down project

It is proposed that HILT Phase II should involve the same mix of participants as HILT Phase I but also involve representatives from the DNER, the RDN, and FE; specifically:

• The Centre for Digital Library Research (CDLR) at Strathclyde University - lead;
• DNER representative
• mda (formerly the Museums Documentation Association);
• National Council on Archives; · National Grid for Learning (NGfL) Scotland;
• Online Computer Library Center (OCLC);
• RDN representative
• FE Representative
• Scottish Library and Information Council (SLIC)
• Scottish University for Industry (SufI)
• UK Office for Library and Information Networking (UKOLN)
• Terminology experts, Alan Gilchrist and Leonard Will (external evaluator)

HILT Phase 2 Pilot

Month:

1

2

3

4

5

6

7

8

9

10

11

12

Activity:

Set up activities

Web-site activities

Service stakeholders e-list

User stakeholders e-list

Community stakeholders e-list

Professional level evaluation process

Scope problem

Identify/monitor evaluation criteria

Identify and monitor subject areas to test

Identify software suppliers

Dissemination activities

Set up focus group dates

Install pilot software

Implement stage 1 pilot (illustrative)

Plan focus groups

Engage with indexing staff

Initial Service Specification

M2M Requirements Study

Detailed Functionality Survey

Hold focus groups

Implement stage 2 pilot service

Terminology map modelling

Full service specification

Estimate and refine full service costs

Interim Report and Feedback

Implement and develop stage 3 pilot

Plan user trials/focus groups

Hold user trials/ focus groups

Improve stage 3 pilot

Compile draft final report

Evaluate project and pilot

Circulate draft final report

Finalise report

Submit/disseminate Report

Project closedown activities

1. Greater understanding of the problem and of the needs of FE and HE users in respect of subject retrieval in the projected JISC Information Environment, both within JISC, JISC services, and - though dissemination activities - in the community as a whole.
2. An in-depth understanding of terminology mapping requirements in the DNER and associated UK services, taking local, regional, national, international, subject-hub, FE and HE, and archives, libraries, museums, and electronic services considerations into account.
3. A working pilot terminologies demonstrator service for the JISC IE (with limited functionality and with a full service possibly requiring a change of software).
4. Requirements, set up and maintenance costs, and costs against benefits, for a future service, including both user and M2M terminological and functional requirements.
5. Final Report on the project, together with appropriate recommendations.

The project will adhere to appropriate standards where these exist and will be advised in this by other participants, such as UKOLN, the DNER, domain representatives and terminology experts. The project is aware of the British standard guide to establishment and development of monolingual thesauri (BS5723:1987) (ISO2788-1986) and the British standard guide to establishment and development of multilingual thesauri (BS6723:1985) (ISO5964-1985) and will consult these and other appropriate works. In addition, it will aim to build UK requirements around terminologies recognised and used internationally (e.g. DDC, LCSH, UNESCO, AAT). In all other matters, the developing DNER/IE standards and the eLib Standards Guidelines would be consulted. The situation regarding other possibly relevant initiatives such as zThes and SRW will be monitored as the project develops.

CAIRNS uses the Z39.50 standard, and the CLD approach used in the SCONE collections database is based on the UKOLN instantiation, and on the original model produced by Michael Heaney when the UKOLN instantiation required extension to take account of practical requirements within SCONE/CAIRNS . The Collections Focus have had a look at these collection descriptions and are happy with them.

Dennis Nicholson, Director, Centre for Digital Library Research, University of Strathclyde