PubMan Func Spec Submission/arXiv mapping

arXiv

Schemas

arXiv currently provides the following arXiv metadata formats via the OAI-PMH interface:

• oai_dc (OAI_DC Schema, Note: parsing is needed see below in mappings )
• arXiv (arXiv Schema, Note: probably a bit outdated as "doi" element is not included in the schema but comes out in the results, see below in mappings)
• arXivRaw (arXivRaw Schema, Note: not complete yet and subject to modification according to schema comments)
• arXivOld (arXivOld Schema, Note: not considered)

For start we will use arXiv metadata format as it seems to require minimum parsing of the metadata values to PubItem.

Mapping from arXiv to PubItem

Arxiv metadata format

1. header/identifier => identifier (without "oai:arXiv.org:" prefix)
(note: this identifier is important because in the output is pointing to the exact version i.e. v1, v2) which is by arXiv used in "citeAs"

2. Authors
2.1. author/keyname =>LastName
2.2 author/forename => Firstname
2.3 author/affiliation => External organization 

3. title => title

4. report-no => Source/sequence-number (only if journal-ref is in, otherwise do not map?)
5. journal-ref => source/title (Parsing not in R3)
6. msc-class => dc:subject
7. abstract => abstract
8. categories => dc:subject
9. doi => dc:identifier (DOI)

10. http://arxiv.org/abs/<header/identifier value> => dc:identifier (OTHER)

Note: missing dates, affiliations are in

OAI_DC metadata format

• Pretty simple as we also use dc metadata in publication profile, but is not correct if not parsed

1. dc:description => dc:abstract (if dc:description does not start with "Comment")
2. dc:date => lists all dates of all versions that exist (earliest date is date when submitted to arXiv, all other are dates when a new revision is done) 
3. dc:identifier => partly is identifier, partly is a source information (journal reference i.e. from journal-ref in the arXiv metadata format)
4. dc:subject => to dc:subject (it is full-name of category ids that are delivered from arXiv format)

===Issues===
*Affiliations: (no possibility for parsing MPI für XXX as organizational units service does not fully support search by organization name) 
**as not certain if we would like to have it within the controlled vocab or directly ask for search-organizations methods from core services an issue is not created as extra requirement for core services. Might be internal requirement for controlled vocab service institutions).
*Parsing of journal names: to check if it is feasible and if possible to relate it in future with controlled vocab service (journals)

Examples

• Record in oai_dc
• Record in arXiv

Open Archives Initiative Protocol for Metadata Harvesting

Digital Object Identifier

Open Archives Initiative

Dublin Core

Max-Planck-Institut