From: Mark T. <mt...@nl...> - 2007-11-20 22:48:22
|
Hi again, I tend to be of the opinion that the more of the record that gets indexed the better, but I'm also slightly unsure about not having the full MARC record around. I'm not a librarian, but I was under the impression that there can be variations between the set of MARC fields/subfields used between institutions, and the way those fields get used. We've certainly already hit cases where we've had to modify VuFind to match our record structure. I guess what worries me is that it seems difficult to define a lossless mapping between MARCXML records and SOLR documents (without just replicating the original MARCXML format). To give a scary example, the way our current Voyager OPAC's record display is configured, there are over 700 tag/subfield combinations that the display processor can pull information from, and that seems like an awful lot of extra fields to add to SOLR :oP Anyway, maybe I'm worried over nothing--I'll be interested to hear other people's comments. For now I guess my vote is to plonk the full MARCXML somewhere (but I don't mind whether that's SOLR or MySQL). We picked MySQL because it seemed slightly easier to us, and it kept a distinction between authoritative data (in MySQL) and indexes that could be wiped and regenerated if need be (in SOLR), but I think SOLR would do the job just fine. Cheers, Mark Andrew Nagy <and...@vi...> writes: > Hello - The previous discussion on importing has sparked my thought on > a new topic. > > We would like to remove the local XML files that are created with the > current import script and store them in a more accessible system. My > initial thought (along with others) is to store the entire marc record > in the SOLR index. This has 2 approaches: create 1 field for the > entire record to be plopped into OR just add any missing fields that > are not being indexed now to complete the marc record. The other > approach is to use a separate database that Mark from down under has > already accomplished. Both approaches have plusses and minuses. > > I am interested in your opinion. My vote is for adding the rest of > the remaining fields to the index so that we have the entire marc > record indexed in SOLR. -- Mark Triggs <mt...@nl...> |