Re: Advanced Keyword Search quirks?

[Doris Munson <dmunson at mail dot ewu dot edu>]

Strange.  The relevancy sort won't work with "wom?n" or "wom*n" alone.  
It will work, however, if another word is added to the search.  "Wom?n 
education" retrieves 968 records when sorted by date or relevance.

The same is true for "go?i."  "Go?i" alone retrieves nothing but "go?i 
and desert" retrieves ten records.

The relevancy search seems to work fine with other words that use a 
wildcard in the middle of the word.  I was able to get results using 
just "colo?r" (46 records), "colo*r" (1343 records), "labo?r" (155 
records), and "labo*r" (10,402 records) as single search terms. 

I'll have to try this again when we upgrade to Millennium 2005 and see 
what happens.

Doris

--
Doris Munson
Systems/Reference Librarian
Eastern Washington University
dmunson at ewu dot edu
(509) 359-6395


Steve Sowder wrote:

> Does relevancy work on your system if the search term contains a "?" wild 
> card character?
>
> Try do a search on "wom?n" and sort the output on relevancy.  On my 2005 
> LE system I get "NO ENTRIES FOUND".  Searching with sor ton date I get 
> 11,197 entries.
>
> Searching for "leader*" produces results how ever it is sorted.
>
> The problem appears to be be when the single wildcard character "?" is 
> used surrounded by other letters. Relevancy sort does not work in that 
> case.
>
> However, another test reveals that "go?i" sort on relevancy just fine. 
> Maybe when their really is more than one character then relevancy sort 
> falls apart.  
>
> To test: think of another test with more than one possible result. We 
> have the classic "wom?n" which does not work. Can anyone think of 
> another?    Sorry, my brain is fried.
>
> Steve
> sowder at andrews dot edu
>
>
> On Tue, 13 Sep 2005, Doris Munson wrote:
>
>  
>
> > The Advanced Keyword Search can be ranked by relevance, if the library 
> > allows relevance sorting.  The library chooses which sorting options it 
> > will allow through the "Advanced Searching: Ranking Options" in OPAC 
> > Options.  The options are:
> >
> > A = Alphabetical
> > D = Date
> > R = Relevance
> >
> > We allow all three, with Date as the default option.
> >
> > By the way, we are on Millennium 2005 LE and relevancy ranking works fine.
> >
> > Regards,
> > Doris
> >
> > Young, Michelle wrote:
> >
> >    
> >
> > > I am relatively new to all of this so I may be off the mark but, It is
> > > my understanding from page #101243 in the manual that the keyword field
> > > is not ranked by relevance and thus, no hits. -Michelle
> > >
> > >
> > >
> > > For the Relevance option, the system uses the following rules to rank
> > > the search results:
> > >
> > > *         The four indexed fields Title, Author, Subject, and Note are
> > > ranked in order. Records that include a particular word or phrase from
> > > the search expression in the Title field will have a higher relevance
> > > than those in which the word or phrase appears in the Author field.
> > > Similarly, records that include the term(s) in the Author field will
> > > have a higher relevance than those in which the term(s) appear in the
> > > Subject field.
> > >
> > > *         For searches on a single word, the system first applies the
> > > "field rule" described above. It then determines relevance based on one
> > > of the following three rules:
> > >
> > > o        Most relevant -- The word is one of the first eight words
> > > indexed for the same record
> > >
> > > o        Next most relevant -- The word is one of the first thirty-two
> > > words indexed for the same record
> > >
> > > o        Last most relevant -- The word is indexed more than once in the
> > > same record
> > >
> > > *         For searches including more than one word, the system first
> > > applies the "field rule" described above. It then determines relevance
> > > based on one of the following three rules:
> > >
> > > o        Most relevant -- The first two words are very near one another
> > > in the index
> > >
> > > o        Next most relevant -- The first two words are near one another
> > > in the index
> > >
> > > o        Last most relevant -- The words are indexed more than once in
> > > the same record
> > >
> > >
> > > Michelle L. Young, MLS               
> > > Virginia Tech, University Libraries
> > > College Librarian, Assistant Professor
> > > myoung at vt dot edu / 540.231.9285
> > >
> > > "The wind of Heaven is that which blows between a horse's ears."
> > > -Arabian Proverb
> > >
> > > -----Original Message-----
> > > From: innopac-bounces at innopacusers dot org
> > > [mailto:innopac-bounces at innopacusers dot org] On Behalf Of Berger, Marianne
> > > C.
> > > Sent: Tuesday, September 13, 2005 2:13 PM
> > > To: IUG INNOPAC List
> > > Subject: RE: Advanced Keyword Search quirks?
> > >
> > > Could you share the number, we have the same thing happening here.
> > >
> > > Marianne
> > >
> > > Marianne Berger
> > > Reference Librarian/Professor
> > > College of DuPage Library
> > > 425 Fawell Blvd.
> > > Glen Ellyn, IL 60136
> > > berger at cdnet dot cod dot edu
> > > Phone: 630-942-2338
> > > Fax: 630-858-8757
> > >
> > >
> > > -----Original Message-----
> > > From: innopac-bounces at innopacusers dot org
> > > [mailto:innopac-bounces at innopacusers dot org] On Behalf Of Steve Sowder
> > > Sent: Tuesday, September 13, 2005 1:02 PM
> > > To: IUG INNOPAC List
> > > Subject: Re: Advanced Keyword Search quirks?
> > >
> > >
> > > Thanks for the info Craig.
> > >
> > > I called the helpdesk and I walked the tech through the search. At the 
> > > NO RECORDS FOUND I heard "huh?".
> > >
> > > I have a call tracking number now.
> > >
> > >
> > > Steve
> > >
> > > On Tue, 13 Sep 2005, Craig Johnson wrote:
> > >
> > >
> > >
> > >      
> > >
> > > > Steve -
> > > >
> > > > Not sure why you're getting the zero "relevance" results - our catalog
> > > >   
> > > >
> > > >        
> > >
> > >
> > >      
> > >
> > > > gives the same numbers regardless of the sort.
> > > >
> > > > According to the manual, "near" is supposed to "retrieve records that 
> > > > contain the specified words or phrases within ten words of each other 
> > > > in the same indexed field." After looking in our catalog and a couple 
> > > > others, it's clear this is not even close to being accurate. The words
> > > >   
> > > >
> > > >        
> > >
> > >
> > >      
> > >
> > > > do not need to be in the same indexed field, nor do they need to be 
> > > > within ten words of each other. "Near" searches yield fewer hits than 
> > > > "and" searches, if not very many fewer, but I can't figure out what 
> > > > the criteria for the different results are.
> > > >
> > > > So... there's a bit more information for you, but I'm afraid the light
> > > >   
> > > >
> > > >        
> > >
> > >
> > >      
> > >
> > > > I'm shedding on your questions isn't really what you're looking for.
> > > >
> > > > Craig Johnson
> > > > Iowa City Public Library
> > > > www.icpl.org 
> > > > ccjohnson at icpl dot org 
> > > >
> > > >   
> > > >
> > > >        
> > > >
> > > > > > > sowder at andrews dot edu 9/13/2005 8:52 AM >>>
> > > > > > >         
> > > > > > >
> > > > > > >              
> > > > We just completed (last Friday) the install of AKS and have since
> > > > discovered some interesting things.
> > > >
> > > > When I do an advanced search using "wom?n" as the search term and 
> > > > "date" as the sort term I get 11,197 hits. When I change the sort to
> > > > "relevance" I 
> > > > get zero hits.  Why?
> > > >
> > > > When I search for "church near school" I get hits that have the word
> > > > "church" in a tag 110 field and "school" in a 500 tag.  The "near" to
> > > > me 
> > > > appears to be a bit "to far" away. Why does it think it is near?
> > > >
> > > > Those who have gone through this before please enlighten me.
> > > >
> > > >
> > > > Thanks
> > > >
> > > >
> > > > Steve Sowder                             Systems Librarian
> > > > sowder at andrews dot edu                      Andrews University   
> > > > --
> > > > This message was distributed through the Innovative Users Group 
> > > > INNOPAC list Public replies:  INNOPAC at innopacusers dot org
> > > > Update your subscription options:
> > > > http://innopacusers.org/mailman/listinfo/innopac
> > > > --
> > > > This message was distributed through the Innovative Users Group
> > > >   
> > > >
> > > >        
> > > INNOPAC list
> > >
> > >
> > >      
> > >
> > > > Public replies:  INNOPAC at innopacusers dot org
> > > > Update your subscription options:
> > > >   
> > > >
> > > >        
> > > http://innopacusers.org/mailman/listinfo/innopac
> > >
> > >
> > >      
> > >
> > > > --
> > > >
> > > >
> > > >   
> > > >
> > > >        
> > > --
> > > This message was distributed through the Innovative Users Group INNOPAC
> > > list Public replies:  INNOPAC at innopacusers dot org Update your subscription
> > > options: http://innopacusers.org/mailman/listinfo/innopac
> > >
> > > --
> > > This message was distributed through the Innovative Users Group INNOPAC
> > > list
> > > Public replies:  INNOPAC at innopacusers dot org
> > > Update your subscription options:
> > > http://innopacusers.org/mailman/listinfo/innopac
> > >
> > > --
> > > This message was distributed through the Innovative Users Group INNOPAC list
> > > Public replies:  INNOPAC at innopacusers dot org
> > > Update your subscription options: http://innopacusers.org/mailman/listinfo/innopac
> > >
> > >
> > >
> > >      
> > --
> > Doris Munson
> > Systems/Reference Librarian
> > Eastern Washington University
> > dmunson at ewu dot edu
> > (509) 359-6395
> >
> >
> >
> > --- StripMime Report -- processed MIME parts ---
> > multipart/alternative
> >  text/plain (text body -- kept)
> >  text/html
> > ---
> > --
> > This message was distributed through the Innovative Users Group INNOPAC list
> > Public replies:  INNOPAC at innopacusers dot org
> > Update your subscription options: http://innopacusers.org/mailman/listinfo/innopac
> >
> >
> > --
> > BEGIN-ANTISPAM-VOTING-LINKS
> > ------------------------------------------------------
> > Teach CanIt if this mail (ID 5089543) is spam:
> > Spam:        http://www.andrews.edu/spam/canit/b.php?c=s&i=5089543&m=8c08de80bed9
> > Not spam:    http://www.andrews.edu/spam/canit/b.php?c=n&i=5089543&m=8c08de80bed9
> > Forget vote: http://www.andrews.edu/spam/canit/b.php?c=f&i=5089543&m=8c08de80bed9
> > ------------------------------------------------------
> > END-ANTISPAM-VOTING-LINKS
> >
> >
> >    
> --
> This message was distributed through the Innovative Users Group INNOPAC list
> Public replies:  INNOPAC at innopacusers dot org
> Update your subscription options: http://innopacusers.org/mailman/listinfo/innopac
>
>  



--- StripMime Report -- processed MIME parts ---
multipart/alternative
 text/plain (text body -- kept)
 text/html
---