From: James F. <jfa...@dr...> - 2008-03-31 16:54:43
|
Andrew, Thanks. All questions (at this point :-) answered. Jim >>> Andrew Nagy <and...@vi...> 03/31/08 12:41 PM >>> > > For 0.8.1, I was planning on tweaking the queries a bit to be more > accurate. > > Could you elaborate? Would you do something other than modify > buildQueryString > to tweak the queries? > ... Well this is kinda hard to elaborate on at the moment since this is what I am working on at this very moment and Im not quite sure of all of the changes that I need to make. Right now I think it should just be the buildQueryString method. All queries *should* be abstracted to the Solr class to allow for backend portability. > > >> * marc2xml2solr.xsl (specifying mappings between MARC tags and > everyday > >> field names); > > >The xsl is no longer used > > OK, thanks. Can you or Wayne verify that the MARC-"every field names" > mappings that > had been there are the same ones that are there by default in the Java > loader? Yes - everything from the PHP importer is in the Java importer. I'm also adding some new fields and they will be added to the new java importer - not the php importer. We hope to have a configurable map file for the 1.0 release. Does this answer your question? > > >The most feedback about the queries that I have had throughout the > lifespan of vufind is to allow a >configurable relevancy ranking system > - so this is something that I very much want to incorporate. Hopefully > >soon :) > > Such a configurable relevane ranking system would, in essence, mean/be > the flexibility to change > the fields and/or their weights that are currently in buildQueryString > in SOLR.php, or would it be something > else entirely? Just the weighting - since we are using defined fields in the schema, all that could be configurable is the weighting. Changing the fields would be a very difficult task since the fields are used throughout vufind. But is something that could be investigated. The Blacklight project uses only dynamic fields. Andrew |