background image
30 DigiCULT
`Facets constitute the major subdivisions of the
AAT hierarchical structure. A facet contains a homo-
geneous class of concepts, the members of which
share characteristics that distinguish them from mem-
bers of other classes. For example, the term mar-
ble refers to a substance used in the creation of art
and architecture, and it is found as a preferred term
(descriptor) in the Materials facet. The term Impres-
sionist denotes a visually distinctive style of art, and
it is listed as a preferred term in the Styles and Peri-
ods facet.
Homogeneous groupings of terminology, or hierar-
chies, are arranged within the seven facets of the AAT.
A broader term provides an immediate class or genus
to a concept, and serves to clarify its meaning. The
narrower term is always a type of, kind of, example
of, or manifestation of its broader context. For exam-
ple, orthographic drawings is the broader context for
plans (drawings) because all plans are orthographic.'
Thus facets (almost always) constitute mutually
exclusive groupings of concepts. Single concepts from
different facets are combined together when index-
ing an object or forming a query. It is a much sim-
pler and more logical organisation than attempting
to form one single hierarchy that encompasses all the
different possible combinations of objects and mate-
rials and agents. Faceted thesauri or classification sys-
tems include the AAT, BLISS
10
and MeSH, and
faceted approaches are being increasingly employed
in Web design.
11
Faceted browsing interfaces to Web databases have
become popular recently. For example, the Flamen-
co system
12
dynamically generates previews of que-
ry results as the user browses different facets. This is
an elegant browsing implementation. However, in
some networked situations, there may be a separa-
tion between terminology services and collection
level services, and query preview may be impracti-
cal where several databases might be involved. In such
circumstances, faceted approaches to searching may
be helpful and this was one of the main aims of the
FACET project.
T
HE
FACET
PROJECT
F
ACET (Faceted Access to Cultural hEritage
Terminology) is a recently completed research
project,
13
funded by the UK Engineering and Phys-
ical Sciences Research Council (EPSRC), which
investigated the potential of the thesaurus in retriev-
al. The project was carried out in collaboration with
the J. Paul Getty Trust, who provided the AAT the
primary thesaurus used in the project, and the UK
National Museum of Science and Industry (NMSI).
An extract of the NMSI Collections Database acted
as a testbed for the project. CHIN (Canadian Herit-
age Information Network) and mda (Museum Doc-
umentation Association) acted as advisors to the
project.
FACET builds on work that started in 1991 when
the University of Glamorgan was commissioned to
develop a hypermedia museum exhibit on local his-
tory from the photographic archives of the Pon-
typridd Historical and Cultural Centre. This inspired
an earlier research project to investigate a query-
based approach to navigation and retrieval, rather
than relying on a priori fixed links. Access routes were
time, space and, as subject index, the Social History
and Industrial Classification.
14
Our aim in FACET was to make use of facet
structure in retrieval. The interface allows users to
select terms from appropriate facets and combine
them in a query. It is possible to conduct highly spe-
cific searches by combining concepts from different
facets. This has the potential for very precise results
if an item in the collection is indexed by these same
terms.
However, in many cases it is unlikely that exactly
the same combination of terms will have been used
in indexing. Perhaps a term has been omitted or the
searcher may have chosen a more specific concept
than the indexer thought appropriate. Alternatively,
there may not be any exactly matching objects in the
collection, although there might be similar objects
potentially of interest.
Toni Petersen, then Director of the Getty Art
and Architecture Thesaurus Project, outlined key
unsolved issues for system designers seeking to take
advantage of the AAT in retrieval (in a discussion of
the National Art Library database at the Victoria and
Albert Museum, London), which was inspirational for
some of our key research aims:
15
`The major problem lies in developing a system
whereby individual parts of subject headings contain-
ing multiple AAT terms are broken apart, individu-
ally exploded hierarchically, and then reintegrated to
answer a query with relevance.'
Our solution involved a technique known as
semantic expansion, whereby search terms are sup-
plemented by additional terms representing similar
concepts, based on their relative positions and rela-
tionships within the thesaurus structure. For exam-
ple, a searcher interested in items made of ebony may
also be interested in items made of a specific type
10
Bliss Classification
Association: Bibliographic
Classification (BC2 or
Bliss), http://www.sid.cam.
ac.uk/bca/bchist.htm
11
Rosenfeld, R. and
Morville, P. 2002:
Information Architecture
for the World Wide Web
(2nd ed.). O'Reilly.
12
Flamenco Search
Interface Project, http://
bailando.sims.berkeley.
edu/flamenco.html
13
FACET project,
http://www.comp.glam.
ac.uk/~FACET/
14
Tudhope, D.:
Geographical Access to
Museum Hypermedia
Exhibits, mda Information,
Vol 2, No 3, http://www.
mda.org.uk/info23t.
htm#Hypermedia
15
Petersen, T. 1994: The
National Art Library and
the AAT. Art and Architecture
Thesaurus Bulletin, 22, 6-8.