Let me briefly introduce myself. My name is Mark Diggory and I am a
software engineer on the "Virtual Data Center" project (
at the Harvard MIT Data Center (, I am also
Harvard's representative to the DDI Alliance ( and
a member of a sub group (the Structural Reform Working Group) mandated
to technically evaluate and restructure the DDI specification. I am
currently evaluating the METS specification with the goal of making sure
modifications made to the DDI provide for interoperability with the METS

Our current discussions for concern the development of DDI 3.0 and
enhancements that would benefit both communities. More specifically, I
am evaluating the avenues for extending METS:

1.) To support "referencing" of DDI instances. Where a METS instance
describes the structure of an entire DDI "Study". In this case METS is
acting only as a Descriptor of the "file" structure of a DDI instance,
in essence, a "Crosswalk" from DDI to METS:

METS                                        DDI

Questions arise concerning how much of a mapping can be produced.

2.) To support the replacement of the archive metadata aspects of the
DDI with extensions in METS or effectively, replacing the "codeBook"
wrapper of a DDI with the METS wrapper, using the Mets Header,
Descriptive and Administrative Metadata sections for content usually
reserved in the stdyDscr and docDscr sections of the DDI. However, I
have questions concerning how the File section and Structural Map could
possibly be utilized to capture the Hierarchical Structure of a DDI
"study". Certainly it could be used to describe the various DDI
instances and their associated files as I outlined in (1). But, could it
also be extended to support describing Variable metadata usually
encapsulated in the DDI? For instance, in DDI, "Structural Metadata"
needs to be more specific to the concept of a data file. The DDI
supports this as descriptors of "Variables" and "Variable Groups" within
the study.

While the structure is somewhat analogous to the <METS:structMap>, there
is considerably much more domain specific metadata that needs to be
captured somewhere, possibly in a dmdSec wrapper.

So ultimately, my thoughts are that such Variable level Metadata could
be encapsulated in a "DDI" extension preserved in the Descriptive
Metadata section. Thus resulting in mappings analogous to


We in the DDI Alliance look forward to any interaction and any feedback
from the METS group on such mappings. We are actively seeking to promote
interoperability between the DDI and other specifications in the Digital
Library Community.

Mark Diggory
Software Developer
Harvard MIT Data Center's VDC Project
DDI Alliance Member