HILT: High Level Thesaurus Project

1
HILT High Level Thesaurus Project

• Investigating the problems of cross-searching
  distributed services by subject in the UK
• Dennis Nicholson, Centre for Digital Library
  Research, Strathclyde University, Scotland

2
HILT is

• A one year project, funded jointly by RSLP and
  JISC
• The main partners are
• CDLR, UKOLN, MDA, NCA, OCLC, NGfL, SLIC, SUfI
• Other stakeholders DNER, RDN, regional
  distributed catalogues

3
Aims

• To study and report on the problem of
  cross-searching and browsing by subject across a
  range of communities, services, and service or
  resource types in the UK given the wide range of
  subject schemes and associated practices in place

4
Looking at

• .Libraries, Museums, Archives, HE, FE, Public
  Libraries, electronic services, bibliographic
  databases, numeric data, and others

5
HILT is charged with

• Researching the problem through literature,
  stakeholders surveys, and focus groups
• Analysing and documenting its exact nature in
  detail
• Determining whether it can be solved and, if so,
• How it can best be solved

6
And with

• Attempting to reach a consensus on the issue
  across the various communities, services and
  initiatives identified by the project as
  stakeholders
• Bearing in mind the need for any viable UK
  solution to be compatible with international
  approaches

7
Aiming for

• A solution that is
• Affordable, sustainable, implementable,
  politically acceptable, useful, future proofed,
  and the rest
• An impossible task, perhaps...
• .but one that has to be tackled

8
Progress to date

• HILT still in the data gathering, analysis, and
  discussion phase of the process
• Initial literature survey complete
• Survey of stakeholders complete
• Various discussions held at project meetings,
  elsewhere (Focus Group)
• And a Literature survey on machine solutions,
  interfaces is underway
• Workshop planned in mid-June

9
Progress to date

• We have also mapped out a two-dimensional
  hypotheses grid that will be used to structure
  and evaluate the data gathered and focus later
  attempts to reach a consensus on a possible
  solution (or a path towards a solution).
• This maps top level and second level options

10
Hypotheses Grid

• Key top level options include
• All adopt a single scheme LCSH, UNESCO, DDC,
  UDC, AAT, New?
• Online service mapping LCSH etc
• Mapping service short term with further
  investigation longer term into whether single
  scheme best

11
Hypotheses Grid

• Key second level options include
• Create new scheme specific sub-thesauri
• Map existing domain specific sub-thesauri
• Multi-lingual capability
• Community control
• Machine assistance, AI
• User mind maps others

12
Stakeholder Survey

• Shows
• LCSH, DDC, UNESCO most common overall
• Archives UNESCO
• Libraries LCSH and DDC
• Museums LCSH, UDC, subject specific thesauri
  like AAT, BMMT
• E-services DDC and LCSH

13
Stakeholder Survey

• Shows
• Mixed approach on handling new editions of
  schemes
• Majority adapt scheme used
• All schemes held to have strengths and weaknesses
• Legacy data a problem

14
Stakeholder Survey

• Shows
• Almost all see subject access and cross-searching
  with other services as important to both users,
  staff
• Mapping between terminologies possible but labour
  intensive?
• Just over half would adopt a single scheme,
  others said depends

15
Focus Group

• People from HILT, Archives, Museums, Libraries,
  E-services
• Participants agreed that
• Cross searching by subject desirable
• Finding solutions to terminology problems
  important
• Problems included resourcing reliance on
  project-based funding legacy material
  differences over standards

16
Focus Group

• Archives
• Subject indexing of collections recent and not
  universal subject indexing to item level a huge
  task.
• Increasing use of UNESCO but also substantial use
  of LCSH.
• Museums
• Subject indexing historically seen as relatively
  unimportant no controlled vocabulary or in-house
  schemes

17
Focus Group

• Museums.
• Library schemes inappropriate for collections of
  objects, so not used
• Resistance to making information available or
  adhering to standards.
• Huge numbers of objects aren't indexed.
• AAT is becoming more accepted and is wider area
  than its name implies
• SHIC (Social History and Industrial
  Classification) used but is not updated.
• MDA maintain lists of object names.

18
Focus Group

• Libraries
• Large amounts of legacy material
• Different versions and local variations of
  schemes, often within one catalogue
• LCSH, DDC most common (but still minority)
• US schemes better maintained, but US slant
• Electronic services
• Metadata contributed by decentralised authors
  with no central co-ordination a problem in some
  DDC and LCSH also

19
Focus Group

• Hypotheses
• Differences mean difficult to make a single
  subject scheme approach work
• Mapping or switching between schemes might work
  would need to investigate user terminology, and
  how mappings are used to create user friendly
  front-end search tools which accommodate the
  complexities of the schemes in use.
• Collection level a minimum starting point?

20
Advice?

• If changing or choosing scheme
• Standard scheme if possible, e.g. LCSH, UNESCO,
  DDC, UDC, AAT
• Preferably in major use in your community (e.g.
  Archives)
• Only adapt if absolutely necessary
• Consulting HILT might help

21
Advice?

• Personal view
• Preferred outcome should be a mapping service
  short term and a longer-term examination of
  whether other options better (Semantic Web
  relevant?)
• DDC backbone? (Multi-lingual including Chinese
  version)

22
Further Information

• Website http//hilt.cdlr.strath.ac.uk/
• e-mail
• d.m.nicholson_at_strath.ac.uk
• susannah.wake_at_strath.ac.uk
• CDLR Http//cdlr.strath.ac.uk/