Overview:
Overview Repository of digitized, historical newspapers
Implemented PREMIS in 2005
To date, applied to ~1.2 million digital objects
What we do:
What we do PREMIS metadata for each file
TIFF
PDF
JP2
ALTO
METS (subset of metadata)
PREMIS metadata is included as <techMD> in METS
What we do - example:
What we do - example <!--PREMIS technical metadata for digital master for page 1-->
<techMD ID="masterPremis1">
<mdWrap MDTYPE="OTHER" OTHERMDTYPE="PREMIS" LABEL="PREMIS object metadata">
<xmlData>
<premis:object>
<premis:objectCharacteristics>
<premis:fixity>
<premis:messageDigestAlgorithm>SHA-1</premis:messageDigestAlgorithm>
<premis:messageDigest><!--SHA-1 checksum (Required)--> </premis:messageDigest>
<premis:messageDigestOriginator>Library of Congress</premis:messageDigestOriginator>
</premis:fixity>
<premis:size>1</premis:size><!--Size of the file in bytes.(Required)-->
<premis:format>
<premis:formatDesignation>
<premis:formatName>image/tiff</premis:formatName>
</premis:formatDesignation>
</premis:format>
What we do – example, continued:
What we do – example, continued <premis:significantProperties>
<ndnp:exemptionSet>
<ndnp:code>TEST1</ndnp:code>
<ndnp:code>TEST2</ndnp:code>
</ndnp:exemptionSet>
</premis:significantProperties>
</premis:objectCharacteristics>
<premis:creatingApplication>
<premis:creatingApplicationName><!--creatingApplicationName (optional)--></premis:creatingApplicationName>
<premis:creatingApplicationVersion><!--creatingApplicationVersion (optional)--></premis:creatingApplicationVersion>
<premis:dateCreatedByApplication>2004-11-22T00:00:00</premis:dateCreatedByApplication><!--dateCreatedByApplication (required for master)-->
</premis:creatingApplication>
</premis:object>
</xmlData>
</mdWrap>
</techMD>
What we do - related:
What we do - related MIX metadata for each image file
Enveloped XML digital signature for METS file
Modified PREMIS schema to omit some mandatory elements:
<objectIdentifier>
<objectCategory>
How we do it:
How we do it PREMIS is generated and inserted into METS by NDNP Validation Library
PREMIS is created by an XSLT transformation of representation information produced by JHOVE
Lessons learned:
Lessons learned Would prefer PREMIS schema to be more modular
Shift in (personal) philosophy: For technical/preservation metadata, less is more:
Collect metadata for immediate needs
Can’t guess what will be needed in the future
If metadata can be derived, then derive when necessary
Lessons learned, continued:
Lessons learned, continued PREMIS as reference model rather than implementation model
Fixities stored in METS/PREMIS require custom tools to verify
Simpler approaches / common utilities more interoperable