There was a question the other day about the status of MADS.  After
reviewing all of the suggestions and discussion on this list pertaining to
MADS, since the last draft, we have a few proposals and several

First, here is a list of changes that we propose for the next MADS draft.

1. Make all of the descriptors allow empty elements, so you can  use an
xlink attribute (URI) in lieu of a  value.

2. Replace <ref> with two elements: <retated type="earlier | later | parent
org | broader | narrower |other">  and <variant type="acronym | abbreviation
| translation | expansion | other">. Both will have xlink attribute.

3. Based on the list discussion pertaining to allowing extensibility of
enumerated lists while preserving interoperability, we propose to adopt
the suggestion to replace all enumerated lists such that the value is a
uri. Lists will be registered externally, with one list to be
distinguished as the "standard" list (which will initially contain the
current values) and a mechanism will be published so that extended or
private lists can be established. See message from Ray Denenberg of June
23, Subject "info URIs for metadata values".

We have the following questions and observations pertaining to suggestions
that we think need further discussion:

1. There has been a suggestion to parse dates so that birth and death dates
can be determined. One approach would be to use ISO 8601 and include a
slash (/) between the start and end date. A second  approach would be to use
a hyphen (-) instead, as prescribed by AACR2. A third approach is to
explicitly distinguish these as elements.

All three approaches present pros and cons, so this needs further
discussion.  If the first or second approach is adopted then there isn't any
change required to the schema (it would be addressed in the guidelines). The
third approach would not be compatible with MODS.

2. It was suggested that we look at EAC DTD for LEAF program and compare it
with MADS.  We haven't had an opportunity to do that (and in fact have not
been able to access the DTD). If anyone would like to do this analysis, we
would welcome it.

3. It was suggested that there be an authority attribute for 2nd level
elements (<name>,  <titleInfo>, <topic>, etc.).   It may not be obvious, but
all these elements already have an authority attribute, via attribute lists.

4. There were questions about intent of MADS examples and whether they
conform to the structure of MADS. We'll try to make the examples more
accurate. One question concerned the lack of an authority attribute in one
of the examples and the possible need to make this attribute required,
since otherwise the purpose of the record is questionable if it isn't
authoritative according to some authority. The example was supposed to
illustrate the case where some institution might want to control headings
locally. However, it probably would be better to make the authority
attribute required and use the value "local" when this is the case.

5  There was discussion of the relationship between MODS and MADS in terms
of the schema definitions.  There are structures in MADS that simply echo
structures in MODS and there were suggestions to develop some sort of type
library for defininitions common to MODS and MADS.  We intend to do this,
but only after the functionality of MADS is more stable; doing this now
would be a waste of effort. We failed to get across the point that the
offending structures are  temporary.

6.  There was discussion of the intent of MADS (browsable strings or term
searching). Our intent was to have a record format in XML that can carry
information about people, institutions, events, subjects, places, etc. One
need is that it be compatible with MARC because of the wealth of data that
already exists in library authority files. I could see, for instance,
using OAIBPH to harvest the data in authority files and have MADS as an
alternative for display of the record. Another need is to have it
compatible with MODS so that you can use them together, perhaps by just
providing a link from the MODS record to the MADS (or MARC) record and
bring in the data from that other source. It was pointed out that MADS
records could be used for term searching so that a smart system might be
able to differentiate 2 names that are the same based on other information
in the record, rather than require the unique browsable string to
differentiate. However, one person pointed out that in a large file with
thousands of records it would be difficult to find a brief display that
would make sense to the user if there are not browsable strings. An
additional suggestion was to include an unparsed element for "authorized
heading" that is separate from the parsed information needed for

These suggestions warrant further discussion. Again, we need to consider
what changes these other approaches would result in for MODS, since we
need to keep them compatible. MODS has become fairly widely used, so there
would be an impact on existing data.

7. There was a suggestion to carry date information without having to know
if it's part of the authorized heading. However it was pointed out that
you need to keep the date associated with the variant or authorized name
since a variant may include a different date from that in the heading.

8. It was suggested to use the ONIX approach for different forms of name
(e.g. name inverted, titlesbeforenames, etc.). This needs further
discussion since it would not be compatible with MODS, and it is unclear
if this much complexity is useful.

9. It was suggested that the schema require attributes authority (see also
above) and relatedType, and that the attribute values be controlled by the
schema. This might facilitate conversion to RDF, however having attribute
values controlled by the schema is not compatible with the suggestion
above to allow URIs for values.

10. There was a suggestion to add an attribute "authorities" to  a
collection as a shortcut to avoid having to define authorities for
individual records. However (if we understand the suggestion correctly)
this would make an individual record dependent on a collection, so that it
couldn't be used as a standalone record. That is, if you extract the record
from the collection it wouldn't have the authority.

11.  A completely alternative approach was suggested to get rid of ref and
refs and have primary and variant names at the same level; also to make
"temporal" a separate element with parsed dates for birth and death (see
above under 7 which points out a disadvantage of this). It isn't clear
whether this approach is so much better than the current one that it would
warrant the extensive changes that would be needed to MODS.

We would like a 2-week period to further discuss these before issuing the
next draft.

^^  Rebecca S. Guenther                                   ^^
^^  Senior Networking and Standards Specialist            ^^
^^  Network Development and MARC Standards Office         ^^
^^  1st and Independence Ave. SE                          ^^
^^  Library of Congress                                   ^^
^^  Washington, DC 20540-4402                             ^^
^^  (202) 707-5092 (voice)    (202) 707-0115 (FAX)        ^^
^^  [log in to unmask]                                          ^^
^^                                                        ^^