| Multi-Lingual Search - Overview | ||
|---|---|---|
| Prev | ||
It will be necessary to nominate, and perhaps develop, a Z39.50 profile to be used in communication between the broker and the data servers. Candidates include the CIMI Profile (see www.cimi.org/standards/index.html#THREE) or a new, simpler profile which would specify much less, leaving ``spaces'' for additional information specific to individual application domains.
For communication between the broker and the metadata repository, a brand new profile would have to be developed, since nothing similar or adaptable currently exists; however, this new profile would be able to make extensive use of the work done on cross-domain metadata in the Z39.50 Attribute Architecture (see lcweb.loc.gov/z3950/agency/attrarch/arch.html) and the utility and cross-domain attribute sets (see lcweb.loc.gov/z3950/agency/attrarch/util.html and www.oclc.org/~levan/docs/crossdomainattributeset.html respectively.) This new profile may well be of benefit to the wider Z39.50 community.
For communication between the broker and the multilingual thesaurus, the obvious candidate is the Zthes profile (see lcweb.loc.gov/z3950/agency/profiles/zthes-03.html) which would perhaps need minor extensions.
Creating a thesaurus, and especially multilingual thesauri, is in itself an extremely demanding task. This can be ameliorated somewhat by limiting the number of supported languages and/or the application domain of the terms.
An effort should be made to locate and use existing thesauri where possible; combining multiple small thesauri into a single large aggregate thesaurus may be a feasible approach.
It may be advantageous to recruit a partner organisation which specialises in terminology control.