Message-Id: <199602292220.QAA29796@library.wustl.edu> Date: Thu, 29 Feb 1996 23:17:43 +0100 From: "Edward C. Zimmermann" <mailto:edz@BSN.COM> Subject: What does proximity mean? To: Multiple recipients of list WEBCAT-L <mailto:WEBCAT-L@WUVMD.WUSTL.EDU>
Hi,With all this talk about proximity matching... Given some thought to the subject: What is an appropriate model for abstract proximity?
While for normal text models-- sentence, paragraph, section, chapter--- the basis for proximity is relatively simple (just apply the semantics for the resp. sieve).. but with generic models this is less clear... While <P> is a common tag for paragraph, its semantics as a representation of a paragraph does not belong to the standard.. In a content model it is not defined..
The question boils down to (using SGML-type tags as an example) deciding what model of proximity is the more "intuitive"--- or maybe "powerfull".
I have been toying with the concept that abstract proximity should not be reflexive but
<TAG1>one two<TAG2>three four<TAG3>five six</TAG3> seven </TAG2> eight </TAG1>
1) (near one two) := TRUE 2) (near two three) := TRUE 4) (near three two) := FALSE 5) (near two five) := TRUE 6) (near three six) := FALSE 7) (near one eight) := TRUE And so on...
Or should "near" be reflexive and only apply in the tag segment: 1) (near one two) := TRUE 2) (near two three) := FALSE 3) (near three four) := FALSE 4) (near two eight) := FALSE
What makes sense?
E. Zimmermann Basis Systeme netzwerk/Germany