The bib Context Set for CQL: Issues, Comments, Questions
See bib Context Set for CQL
- Can titleUniform, titleAbbreviated, titleSeries,
etc. be expressed as modifiers to dc.title?
There are two independent dimensions to be represented, "type" and portion.
Type:
Abbreviated, Uniform, Translated
, Alternative, Series
Portion:
main, sub, partNum, partName
Logically the first is an index modifier and the second is a relation modifier. But unfortunately there is no such thing as an index modifier. (See 3.) So that would mean overloading the relation modifier with information that is logically associated more with the index than the relation. We could do that, but would that really reduce complexity, versus defining separate indexes?
There is also sentiment that separate indexes is more logically sound than a single index with many modifiers.
Finally, there is some resistance to mixing dc indexes with bib qualifiers.
- Same question for subject. Can bib.subjectPlace, bib.subjectTitle, bib.subjectName, bib.subjectOccupation be expressed as modifiers to dc.subject. Also, dates.
- Why is there no such thing as an index modifier (and if there isn't why not invent it)?
You can't say:
title/type=abbreviated=/portion=main dog
This construct is syntactically ambiguous, because you don't know where the index modifier ends and the relation begins. So we considered the possibility of defining index modifiers but not allowing them the flexibility of relation assignments, thus for example:
title/abbreviated=/portion=main dog
Where title/abreviated implicitly means title/type=abbreviated. So the suggestion, in effect, was to designate one single modifier type for an index which would be implicit, and then the actual value of that type would appear as the modifier. This approach seemed overly complicated, as well as overly restrictive.
Or another possibility, using a different character than '=' within the index relation, for example '*'. This approach was rejected because we were running out of special character, and in any case it played havoc with the BNF.
Another, completelely separate issue, was tacking modifiers onto DC indexes. The DC set, representing the 15 DC elements as indexes, never has raised any controversy (even though Dublin Core has never defined such a thing) because it's just the simple 15 DC indexes, with the DC semantics. Adding bibliographic modifiers might raise some controversy. Then again maybe not, anyway this was a secondary issue.
- Could all of the 'authority' modifiers be merged into just
bib.authority?
For discussion.
- Is the purpose of bib.name to search for a name anywhere in the record
(could be dc.creator, could be dc.subject, could be dc.contributor,
etc.)?
Anywhere in the record that the server thinks is supposed to be a name. However:
- Note the separate index bib.subjectName. Thus bib.name would not be used to search for a name subject.
- bib.name might apply to the whole record (or most of it) when unqualified. However, it may be qualified by 'role'. Thus if role=author, then it would be restricted to searching for authors.
- Given the profusion of national bibliographical identifier schemes, making bib.identifierAuthority have a limited vocabulary is not going to scale, and it would be useful to allow ad hoc registration of new authorities.
Done. Introduced extensibility mechanism for identifier types.
- Journals often have "New series" for their issue numbers, indicating they've started counting from 1 again; the current scheme does not indicate re-numbering.
Add modifier 'renumber' whose value is an integer. If omitted, zero is assumed (no renumbering). 1 means that numbering has been restarted once, etc. e.g.
bib.titleSeries=/bib.portion=partNum/renumber=1 nnn
- Subject classes are restricted to Place, Title, Name and Occupation. Why not for instance Curriculum Objective, or chronological terms?
For discussion.
|
