Page 1 of 1

dash prefix misbehaving

Posted: Fri Aug 30, 2002 1:33 pm
by bill.marel
We have data for Martin Luther and for Martin Luther King Jr.

When searching for Martin Luther -King we would expect to get hits on Martin Luther and no hits on Martin Luther King Jr. This is not the case; the highest ranked hits are all of Martin Luther King Jr. However, we have discovered that Martin Luther -(King) gives the desired results of hits on Martin Luther only.

When we try similar searches using the dash prefix, they seem to work. An example that works is Tennis Martina -Hingis does not retrieve Martina Hingis but does retreive Martina Navratilova.

King is not in our list of noise words.

We have:
<apicp keepnoise on>
<apicp alpostproc 1>
<apicp alintersects 1>
<apicp minwordlen 4>
<$suffixes = s>
<apicp suffix $suffixes>
<apicp defsuffrm 0>
<apicp keepeqvs 1>
<apicp alequivs 1>
set likepallmatch = 1

dash prefix misbehaving

Posted: Fri Aug 30, 2002 2:28 pm
by Kai
What's the version of texis you're using (date printed by texis -version)?

dash prefix misbehaving

Posted: Tue Sep 03, 2002 2:59 pm
by bill.marel
Texis Web Script (Vortex) Copyright (c) 1996-2001 Thunderstone - EPI, Inc.
Commercial Version 4.00.1004732495 of Nov 2, 2001 (i686-unknown-linux2.4.2-64)

dash prefix misbehaving

Posted: Tue Sep 03, 2002 3:06 pm
by bill.marel
Also we have done some more testing to determine if there is anything special in our system re the word King. billie jean -king and billie jean -(king) both correctly exclude hits on Billie Jean King the tennis player. martin luther -king still doesn't work and we don't know why.

dash prefix misbehaving

Posted: Tue Sep 03, 2002 4:11 pm
by mark
We've not been able to replicate that behavior here with that (or any) version.

There should be no difference between -king and -(king) except that the former will do equiv lookup with your settings. -~king should behave similarly to -(king) in your case. Do you have any custom equiv entries for "king" or containing "king"?

You mention the noise list. Do you have a custom noise list?