[Koha] Question on Z39.50 cross-format queries

[Manos Petridis]

>________________________________
> From: Robin Sheat <robin at catalyst.net.nz>
>To: koha at lists.katipo.co.nz 
>Sent: Tuesday, August 14, 2012 1:07 PM
>Subject: Re: [Koha] Question on Z39.50 cross-format queries
> 
>
>Op 13-08-12 21:36, Manos Petridis schreef:
>> a) It it were simply a case of either/or, then why can I select the MARC flavour of the various Z39.50 origins set up at my installation? What is the reason to let koha know of the MARC flavour/dialect, if it not able to use the data that each server sends? There must be some functionality behind the dropbox selections, even if it only defines the query statement dialect.
>
>Koha needs to know what you are using, so that it knows what field to
>pull from for the title, and how to index the title in zebra to make it
>correctly searchable, for example. There are a number of other cases
>where the differences are a bit deeper than just different field IDs.
>
 
That's exactly my point! Koha knows 
a) what MARC flavor my installation is using 
b) what MARC form each Z39.50 origin accepts requests and supplies its results in
It should therefore be able to know "what field to pull from for the title" etc. 
 
I would expect that zebra indexes the various koha tables, not the imported MARC records, i.e. works on data past the mapping phase.

>>>b) From What I can see in the data presented (in the MARC form window) it sometimes is just a matter of mapping. For example, it seems that 
>
>It's not really feasible to correctly map from UNIMARC to MARC21 in an
>automated fashion. MARC rules extend beyond what field something is in,
>to what information is made available, and how it is formatted within
>the field. You could probably get a pretty good automated conversion,
>but it would almost certainly be lossy.
>
 
If I understand things correctly, Z39.50 look-ups are used in koha cataloging as an aid, rather than a gospel to be strictly adhered to. 
This is evident both in the differences between MARC records from different reputable origins - apparently all manned by skilled librarians - and by the nature of some of the information provided in query results: why do I care for example where a specific item is stored in library X, when I'm only interested to document my own copy in my own library? Lossy conversion can be a boon in this context.
 
As long as standards are well defined, one can perform formatting conversions correctly (character to numerical, left/right justified, etc.), partially at least, i.e. for specific common fields. Alas, I'm perl-impaired so I'm in no position to substantiate this view.
 
>
>-- 
>Robin Sheat
>Catalyst IT Ltd.
>✆ +64 4 803 2204
>GPG: 5957 6D23 8B16 EFAB FEF8  7175 14D3 6485 A99C EB6D
>_______________________________________________
>Koha mailing list  http://koha-community.org/
>Koha at lists.katipo.co.nz
>http://lists.katipo.co.nz/mailman/listinfo/koha
>