From: Tania F. <ta...@br...> - 2009-12-22 19:01:21
|
Demia, Thanks for the preview of the delights that await us in rc2. We are definitely looking forward to rc2, but this close to our launch date it's not possible for us to upgrade. I'm really just trying to put together a coherent explanation for staff who are used to traditional catalogs, as to why this behavior is to be desired. If not specifically the OR vs AND, then the "get a lot and narrow it afterward" philosophy of nextgen catalogs in general. Even with the rc2 improvements, I am sure I will need to keep the explanation handy, since it's a significant departure from the behavior of traditional opacs. I guess what I'm saying is I need the philosophical argument(s) about why nextgen catalogs do searching this way, and philosophical whys are not my strong suit! :-) -- Tania Fersenheim Manager of Library Systems Brandeis University Library and Technology Services 415 South Street, (MS 017/P.O. Box 549110) Waltham, MA 02454-9110 Phone: 781.736.4698 Fax: 781.736.4577 email: ta...@br... > -----Original Message----- > From: Demian Katz [mailto:dem...@vi...] > Sent: Tuesday, December 22, 2009 1:51 PM > To: Tania Fersenheim; vuf...@li... > Subject: RE: [VuFind-General] VuFind searching: questions > about OR vs AND > > VuFind's default search behavior has been improved > considerably in 1.0RC2. As you say, 1.0RC1 uses OR searches > by default since it captures the most records, and relevance > ranking helps bring the good stuff to the top. However, as > I'm sure you've noticed, this brings with it a certain > potential for getting misleading or completely irrelevant results. > > RC2 improves on the situation in two ways: > > 1.) It uses Solr's Dismax search handler whenever possible. > Dismax allows searching across multiple index fields with > different weights more easily than the old Lucene-based code. > It also has a useful setting called "Minimum 'Should' Match" > (mm) which allows you to specify very detailed preferences > about how many search terms need to be matched in order to > include a record in search results. This allows you to do > things like force a two-word search to match both terms, but > allow a three-term search to return hits that match either > two or three of the terms. It takes trial and error to find > the setup that works best for you, but when you find it, you > can have the benefits of the OR search (inexact matches) > without the downside (garbage that only matches on one out of > many search terms). Out of the box, RC2 is configured to > require 100% matches -- so it works just like your old OPAC's > AND search -- but you can easily add the mm setting either > globally or on a search by search basis (in case you want to > configure, say, Authors differently from Subjects). > > 2.) There are some advanced search cases where Dismax can't > be used due to some technical limitations. In these cases, > VuFind may still revert back to a Lucene-based OR search like > it used in the old days. However, a new configuration file > allows you to customize exactly how searches are constructed > in the absence of Dismax, so you can eliminate the OR option > if you don't like it. > > For all the technical details about how to configure this > stuff, see this page in the Wiki: > > http://vufind.org/wiki/searches_customizing_tuning_adding > > If you have any further questions, I'll be happy to answer them! > > - Demian > > > -----Original Message----- > > From: Tania Fersenheim [mailto:ta...@br...] > > Sent: Tuesday, December 22, 2009 1:25 PM > > To: vuf...@li... > > Subject: [VuFind-General] VuFind searching: questions about > OR vs AND > > > > > > VuFind does an OR search by default (e.g. if I enter an > Author search > > for > > william shakespeare the search as executed is author = william OR > > author = > > shakespeare). > > > > Our old Opac does an AND by default - I would guess most existing > > catalogs > > do an AND. > > > > Can anyone explain the Pros of having VuFind search this way? > > > > One possible Pro we have come up with is that you want to > get as much > > back > > as possible from a search and let relevance ranking sort > the results. > > > > > > -- > > Tania Fersenheim > > Manager of Library Systems > > > > Brandeis University > > Library and Technology Services > > > > 415 South Street, (MS 017/P.O. Box 549110) > > Waltham, MA 02454-9110 > > Phone: 781.736.4698 > > Fax: 781.736.4577 > > email: ta...@br... > > > > > -------------------------------------------------------------- > --------- > > ------- > > This SF.Net email is sponsored by the Verizon Developer Community > > Take advantage of Verizon's best-in-class app development support > > A streamlined, 14 day to market process makes app > distribution fast and > > easy > > Join now and get one step closer to millions of Verizon customers > > http://p.sf.net/sfu/verizon-dev2dev > > _______________________________________________ > > VuFind-General mailing list > > VuF...@li... > > https://lists.sourceforge.net/lists/listinfo/vufind-general > |