linkingmetadata
BHL Metadata for cross-system linking
1. How do we get our ILS records updated as stuff is scanned?
2. How do we ensure we can provide an up-to-date list of resources for link resolvers?
3. How do we update serials (and mono?) lists to know what stuff has been done as opposed to
being on the "to do" list?
Initial thoughts:
We need a one-stop shop from BHL which contains all the relevant fields we may need for each
of the above. Spreadsheet of possible mappings attached. The rec_key to ILS may be hardest!
BHLmetadata.xls
SFX Issues:
- If we can aggregate "vols per title", this will be the best fit for SFX. It would require a splash page link at title level to link to.
- Tech Note: BHL currently supports aggregating items into a title:
- Journal of the Royal Institution of Great Britain
- SFX needs an object id. Doesn't have to be ISXN. However if we don't provide an ID in popular use in source dbs (ie ISXN), uptake may be less.
- Tech Note:
- BHL creates a TitleID for each title, such as 2258 in the example above. We have created a series of "Title Resolver" tables that xrefs our BHL TitleID with other title level identifiers, such as:
- ISSN (relatively few in our collection thus far)
- ISBN (ditto)
- OCLC
- This list can be augmented as needed to include things like TL2 numbers, BPH abbreviations, DOIs, LSID, etc., as a way of linking
- These are Key-Value pairs, with 'OCLC' being a key and '02917176' its value. Example:
- BHL needs to provide this xref in an indexable form.
- What do partner ILS require?
- What do other library ILS require? Is it different?
- What do metadata aggregators require?
- What method(s)?
- RSS
- Atom
- CSV
- NLM-based custom service
- DISCUSSION BELOW - Please continue it there!
- Should we consider retro assigned ISSNs at title level as a matter of urgency to assist?
- Tech Note: I think we've determined that we can't yet assign ISSNs to our electronic titles.
- Note from Tom Garnett: We will have ISBN's. Would ISSN's help? I am in discussion with Bowker and am willing to make a pitch.
- Note from Chris to Tom: ISSNs are not assigned by Bowker.
As a subsequent step, we may be able to accept BHL vol level data if we can create a
parser to take requested vol/date and append it to a lower level id (which may have to
sit in a separate SFX or even external table.)
Comments/suggestions??
10 Jul 2008
We've been monkeying around at the NHM with BHL metadata drawn from the IA and experiementing with the sort of thing we could do to aggregate it. See
http://www.nhm.ac.uk/library/bhlseriallist/index.php/pages/bhlmetadata
What, exactly, the Portal needs to provide to libraries
Please add to list below as Chris needs this info !!
Purpose: Regularly scheduled export of Titles & Items available through BHL portal.
Title-level Information: (e.g.
The Journal of Shellfish Research link)
-
Sample download
BHL TitleID
OCLC
ISSN
ISBN
LCCN
Title (MARC 245) _meta.xml.<title>
Publication info (MARC 260)
Local Library ID <title-id> in _meta.xml
- from MARC record or WonderFetch
- NHM _metasource.xml.<cclquery> OR _meta.xml.<title_id> OR _marc.xml.<controlfield tag="001"> with "Catkey" present
Scanning Library
Item-level Information: (e.g. 'v.2' of
The Journal of Shellfish Research link)
-
Sample download
BHL ItemID
IA Identifier
Local Library ID (Barcode, other): <identifier-bib> in _meta.xml
Vol from/vol to _meta.xml.<volume>
Year from/year to _meta.xml.<year>
Format:
csv preferred (NHM)
key/val lookups for OCLC, ISSN, normalized title???, BHL titleid, BHL itemid
Responses to Version 1 release:
1. start yr/end yr are in title.txt and not item. Where are you pulling that info from? If
it is an accurate reflection of scanned dates, great. If not, should we have the date/year
fields in the item level rec (alongside vol) which goes via wonderfetch? We also need accurate
holdings summaries for systems like SFX (if possible).
CDF: Start/End for the Title come from MARC Leader, so is as accurate as the home library's cataloging. I don't think we should include dates from the item-level record(s) here since we're talking about titles. Can you describe more of what you need for SFX re: holdings? What format? Is a text string enough? Can we generate that from the items scanned?
2. Missing fields, OCLC, ISSN, ISBN, Scanning institution
CDF: Yes, will include those in the next revision, which should be out by 7/25/08.
3. Possibly useful to have an 001 and 035 field from Marc (there may be multiple 001s, so maybe
just first found)? Quite a few ILSs have record key in there which could be a second option
and for where wonderfetch was not in use.
CDF: Will have to think on this one. We're not producing this list from the MARC directly; we hoover in certain pieces of MARC into the BHL db to facilitate search & browse and we're not specifically pulling in 001 & 035. Will check on this.
4. Should we consider allowing short-title based openurls? May not be 100% accurate, but in
some cases maybe the only link people have from a citation in older material.
e.g
http://www.biodiversitylibrary.org/titleshort/Traite%20des%20arbres%20forestiers
CDF: We can already link out to a fuzzy title search like this:
http://www.biodiversitylibrary.org/Search.aspx?searchTerm=Traite%20des%20arbres%20forestiers
5. Short title should appear in title.txt for ref purposes and quick import to knowledge
bases.
CDF: Just the short title (245a) and not the full title, or do you want the short title + the full title?
