Talk:PubMan File Properties

Component properties and/or file metadata record?[edit]

Resources[edit]

file properties which are defined and maintained by Framework: http://www.escidoc-project.de/schemas/soap/item/0.6/components.xsd
escidoc application profile file: ESciDoc_Application_Profile_File

Discussion[edit]

The file-size will be removed from the properties - that means we need to create it in our file metadata record. But on the other hand side, keeping only file-size in the metadata record as a single metadata makes no sense to me. So the questions are:

shall we stick with the file size only in the metadata record
shall we put the file size and redundantly keep as well properties in the metadata record (may be important for things like LTA, dissemination)
shall we remove the file size completely from the metadata record

Are there any use cases in which the framework change a component property? Something like "creation-date", "created-by" and "valid-status" is probably completely controlled by the framework while other properties like "description", "content-category" are only modified by the application only (Am I right?). So, we probably don't want to copy the complete set of properties, but duplicate selected information (= all terms described by functional experts = listed in application profile file) --Inga 11:46, 15 May 2008 (CEST)

exactly, and we should communicate it to FIZ - by defining our metadata schema (most of it is duplication of selected properties with exception of file-size) --Natasa 14:02, 15 May 2008 (CEST)

For items we already split information between metadata record and property (e.g. the item status/identifier) is in the properties while descriptive information is stored in the metadata record. This would be fine for files too - if we have a common understanding about what characterize a property and what characterize an attribute in the metadata (see below) --Inga 11:31, 15 May 2008 (CEST)

as old story as the world :)) ... so far i think we developed some common understanding on what is property and what is metadata (property is used for FW internal functions. We can duplicate it in the MD record or not, but if we duplicate, we should make sure from our (client) side to make them in sync - at least for properties that we are providing to the FW. To clarify:

there are properties that FW uses and are provided by us
there are properties that FW sets-up (they are read-only) automatically and we use them --Natasa 14:02, 15 May 2008 (CEST)

The definition "properties are whatever information is required by framework to handle object by internal functions" is indeed a bit other notion than I had till today. The distinction may be clear from the framework developer point of view, but is not necessarily transparent for the framework-external user/developer. --Inga 14:47, 15 May 2008 (CEST)

Outcome[edit]

Agreed to duplicate all properties/terms functionally required to describe a file (= ESciDoc_Application_Profile_File) to component metadata and component property.

filename (mandatory, 1) : http://purl.org/dc/terms/title
content-category (mandatory, 1): http://purl.org/escidoc/metadata/terms/contentcategory
file-description (optional, 0-1): http://purl.org/dc/terms/description
internal-file-id (mandatory, 1) : http://purl.org/escidoc/metadata/terms/identifier
file-PID (optional, 0-1) : http://purl.org/dc/terms/identifier
file-size (mandatory if not external-locator, 1) : http://purl.org/dc/terms/extent (TODO: Clarify with FIZ)
external-locator (optional, 0-1) : http://purl.org/escidoc/metadata/terms/locator
MIME-Type (mandatory if not external-locator, 1): http://purl.org/dc/terms/format

Metadata schema available at:<tobedefined>

Further questions[edit]

Could somebody please explain why the file size should be removed from properties, thus which characteristic makes an information to a property? --Inga 11:31, 15 May 2008 (CEST)

Not sure MPDL can explain, it's FIZ decision, as the file size is calculated from the OS for the storage to my understanding and in different size units

Okay, than I understand that your question from above "shall we remove the file size completely from the metadata record" is rather a technical than a functional question. If file size can automatically be calculated when displaying file information on a pubman record page this would be fine as well :) --Inga 16:10, 15 May 2008 (CEST)

I agree;-)--Ulla 11:22, 26 May 2008 (CEST)

How would the object look like? Item xsd with one metadata record and additional metadata records for each component? --Inga 11:51, 15 May 2008 (CEST)

Item.xml only contains links to component object and one or more metadata records (only one metadata record is of type "eSciDoc" and this one is used by Lucene for indexing e.g. VIRR object has MAB and eSciDoc metadata record).

Component.xml contains the metadata records (as much as there are). Besides eSciDoc metadata (from Application profile) this contains also technical metadata record (generated from JHOVE)--Natasa 14:05, 15 May 2008 (CEST)

IMHO, we need to cross-check current available information anyway as soon as we have to deal with PREMIS compatibility...right?--Ulla 11:25, 26 May 2008 (CEST)

Mapping from component properties to pubman metadata[edit]

for build 0.159, schema location at components schema