Support call: metadata advice

Institution: Bangor University
Date: 28th April 2009
Subject: Metadata

Bangor University are using their master publications database to feed information about research outputs into their repository. They recently called the WRN team to seek advice about transferring various fields of metadata from their database into the repository, and specifically which of the Dublin Core (DC) fields were most appropriate for various bits of publication information. These are the fields in question:

Conference name
We advised to map this to the publisher field in DC. Even if the conference is not strictly 'published' this is probably still the best place for the conference name information to be stored as the DC qualifier nameConference can be used in this field.

Commissioning bodies
We advised to map this information to a DC description field. It could be mapped to a contributor field but the catch all of description is probably the better option.

Patent number
We advised to map this item to a DC identifier field. At present there is no qualifier within the identifier field for patent numbers but one could be used in the metadata scheme as a local modification so on the full item record the field would be tagged as identifier:patent number.

Media of output
During discussion we discovered that the media of output field is used at Bangor to house a mixture of information, sometimes containing information about the type of deposit such a conference poster and sometimes about the format it comes in such as on CDROM etc. As such we advised to again map this item to a DC description field as it is a catch all place for information to go, much like an additional information field. The varying contents of this field raised an interesting question about item types in the repository. The type information about publications in the Bangor publications database is stored as a letter code in a specific field and this should transfer over to the repository and into the DC type field quite happily. However, when we were asked about agreed terminology for the types within the repository we uncovered an interesting conundrum. There is a recommendation within DC standard to use a vocabulary within the type field, but this is quite restrictive and does not align with what are typically considered to be publication types in the repository. The vocabulary categorises most repository item types as 'text' which is not sufficiently detailed to indicate whether the item is a journal article, a book chapter, a report etc. We consequently advised Bangor that two DC type fields can be used within the metadata schema, one to store the required DC vocabulary terminology, and a second one to contain the more descriptive item types we would expect to see. At Aberystwyth University we have made this addition to our metadata scheme and added a local qualifier to the second type field - type:publicationtype - and defined our own set of values for this field to reflect the various types of information we store.

If anyone would like further clarification of this information, or would like help with their own metadata choices then please do not hesitate to contact the WRN team via wrnstaff@aber.ac.uk