Re: Facet algebra

Dear Aida,

Thanks a lot for explaining the terminology used in your area by 
giving specific examples. This allows me to answer your questions. Of 
course my answers come from a Computer Science (CS) point-of-view:
----
1/ Facet Citation Order Rule.
   If I understood correctly, this rule mostly concerns object 
indexing.  Within object indexing  I usually consider (e.g. see my 
paper in ER'2003) indexing statements of the form:   index(1)={Crete, 
SeaSports} where 1 denotes a resource (e.g. a book, a Web page, etc). 
The indexing statements could be also ordered, i.e. index(1)=<Crete, 
SeaSports>, and the  ordering of the terms could be taken into 
account during query answering.  Of course this decision depends on 
the application. In any case, the underlying technical issues (for 
query answering) are trivial from a CS point-of-view. 


2/ Filing Order Rules.
Again the order does not play a critical role  from a CS point-of-
view.  
However you mentioned that there are situations in knowledge 
classification where some generally applicable facet is not to be 
used (e.g. the  facet of materials is not applicable for chemistry).  
The definition and management of this kind of rules,   is according 
to my opinion the most challenging technical issue and the Compound 
Terms Composition Algebra gives a solution to this issue. The algebra 
actually facilitates the definition of this kind of rules by giving 
the opportunity to the designer to use both "positive" and "negative" 
rules and by providing a method for inferring new rules  from the 
given (by exploiting the structure and the semantic interpretation 
that is given to the hierarchies).

3/ Restriction on Synthesis and Rules on Combination Between Facets.
Without loss of generality we can view RELATORS as terms  and we can 
organize them hierarchically. This means that the  Compound Terms 
Composition Algebra can give a solution to this problem as well, i.e. 
it can make the definition of this kind of rules easy, flexible and 
sharable.
----
    
At last, I would like to say that we shall not confuse XFML with  the 
Compound Terms Composition Algebra. XFML is an XML format for 
exchanging faceted metadata,  while the Compound Terms  Composition 
Algebra  is an algebra that can have several applications in 
knowledge representation and reasoning. One possible application is 
in faceted classification.  I extended XFML to XFML+ just for 
demonstrating how the Compound Terms Composition Algebra could be 
applied in the Web context. For instance, this extension allows us to 
exchange expressions that define restrictions on synthesis.

Best regards,

Yannis


--- In [email protected], "Aida Slavic" 
<aida@a...> wrote:
> [!long text with examples, please skip if not interested in 
citation/filing
> order
> of facets in knowledge organization classifications]
> 
> Yannis,
> I think this is a great work you've done.
> I have had looks at the articles and I have the feeling that this
> may be the answer but I need to spend more time on this to bridge
> the gap between different perspectives in viewing facetted 
vocabulary.
> 
> When XFML was first published I found it too thin to support binding
> of facetted vocabulary in a way that it is used for knowledge 
organization
> (KO).
> Now when I have my vocabulary in  a proper database I would like to
> be able to do export in a more transportable format such as XFML
> without loosing too much data. I will have look at XFML+
> 
> KO classifications I work on is based on the principles of facetted
> (analytico synthetic
> classification).
> Vocabulary organized in facets contains subfacets and hierarchically
> structured concepts
> There are also facets that can be combined without limitations and 
facets
> that have limited
> area of application. Also the compound concepts themselves can be
> synthesised using
> relationship indicators (related to, influenced by, expanded with 
etc.) All
> together there
> are three levels of synthesis or compounds building.
> 
> The main goal of every knowledge classification is to be able to
> display/present knowledge
> from general to specific. Traditionally this is achieved through 
both the
> definition of facets
> and the rules these have to be combined and presented/filed.
> 
> Here is very simple example.
> 
> Literature is organized into main facet according to the language 
literature
> is written in.
> So I have  English, French, German literature (you can also call it 
the
> facet of language
> in the literature)
> In the literature I also have three other facets that are 
applicable ONLY in
> the field of
> literature
> 
> facet of theory [criticism, stillistics, philological studies]
> facet of time/period (old, middle, modern)
> facet of genre (poetry, drama, fiction, esseys,correspondence etc.)
> 
> 1) first level of synthesis (1st rule of combination)
> Any concept in literure can be built using above mentioned facets
> [English][Old][Poetry]
> The order of these facets is usually defined by rule of combination 
that
> will help achieving
> general->specific presentation by the virtue of horizontally citing 
first
> what is the most specific
> to the subject, and last: what is the least specific to the subject.
> So you cite first subject itself [English literature] then time 
facet [Old]
> as it determines
> the subject closer than the facet of form [genre]
> Thus when filing these combinations you will have
> [English][Old][Poetry]
> [English][Old][Drama]
> [English][Old][Fiction]
> [English][Middle][Poetry]
> [English][Middle][Drama]
> [English][Middle][Fiction]
> [English][Modern][Poetry]
> and NOT
> [English][Poetry][Old]
> [English][Poetry][Middle]
> [English][Poetry][Modern]
> [English][Drama][Modern]etc.
> Presentation of some disciplines and fields is not very much 
disturbed by
> order of facets.
> This is the case here. As it is well obvious, one may want to have 
different
> order for different purpose
> of gathering - but this  makes 'managing' of rules of combination 
even more
> important
> (for more intellectually demanding example see the following
> http://www.ucl.ac.uk/fatks/religion.htm)
> 
> 2nd level of synthesis
> 
> Every modren KO classification keeps generally applicable 
vocabularies such
> as place, time, persons, languages
> in a separate facets. For instance concept such as 'children' can 
be used
> accross the knowledge structure
> (Literature for children, games for children, children cloths, child
> abuse).These facets have now limitation
> of use and synthesis (almost), they can be combined among 
themselves and
> with any subject concept.
> 
> These are for instance generally applicable facets
> 
> Facet of Persons (over 200 concepts) (e.g. see outline structure at
> http://www.ucl.ac.uk/fatks/o_person.htm)
> Facet of Place (over 700)
> Facet of Time (over 700 concepts)
> Facet of Form( over 400)
> Facet of Language (over 1000)
> 
> (NB these facets have subfacets and hierarchically organized 
concepts as you
> can see if you follow link
> above)
> 
> When you have a subject which you want to specify further it works 
like
> this:
> 
> 			[persons]     [time]           [place]       
[form]	[language]
> English literature[for children][in 19th century][in Australia]
[study]  [in
> French language]
> Employment		[children]    [20th century]   [USA]	   
[report] [in English
> language]
> ----------->>>>>>----------decreasing
> specificity---------------------------------->>>>
> 
> The way you order/cite these facets horizonatally is in decreasing
> specificity/concretness
> Explanation: while 'persons' very much define the subject as well 
as time,
> the 'form' and the 'language' in which
> the document is written, cited last, are the least relevant for the 
subject
> itself.
> 
> When filing compound terms you have the reversed rule (of increasing
> specificity):
> 
> English literature [in some language]
> English literature [in some language][in some country]
> English literature [in some country]
> English literature [in some country][in some time]
> English literature [in some country][in some time][for some kind of 
persons]
> English literature [in some time]
> English literature [in some time][in some country]
> English literature [in some time][in some country][for some kind of 
persons]
> English literature [for some kind of persons]
> 
> 
> But there are situations in knowledge classification where some 
generally
> applicable facet
> is not to be used. For instance the part of facet of materials is 
irrelevant
> for chemistry where
> 'materials' is the actual subject (e.g metals), or some part of 
facet of
> 'properties'(properties
> of action and movement e.g. velocity, mass) is not relevant for 
physics.
> These things are
> usally documented in existing classification as instruction note, or
> application which is a simple text.
> 
> 3) 3rd level of synthesis
> 
> Is when you want to relate two or more distant subject facets 
(compound or
> simple)with 'relators'
> 
> English literature <RELATION> Religion
> English literature [in some country]<COMPARISON> French literature 
[in some
> century]
> Religion <INFLUENCE>English literature
> 
> Please not that classification uses notation (symbols) instead of 
the terms
> I have used here
> - and thus preserve the differnce between concepts and terms.
> 
> These bunch of rules makes the use of big knowledge classification 
harder to
> grasp and
> they are usually very poorly structured and harder to record, 
document and
> exchange in machine
> readable form.
> Anyway if you look at the data model for faceted classification I 
am working
> with
> you may better understand the complexity of this
> http://www.ucl.ac.uk/fatks/datastructure.doc
> 
> Do you think this can translate into XFML?
> 
> Aida
> 
> aida
> 
> -----Original Message-----
> From: yannis_tzitzikas [mailto:tzitzik@i...]
> Sent: 03 September 2003 08:35
> To: [email protected]
> Subject: [facetedclassification] Re: Facet algebra
> 
> 
> Dear Aida,
> 
>    My area is Computer Science and thus I am not very familiar with
> the terminology used in library and information science. However I
> would really like to know how people from your area view the 
Compound
> Terms Composition Algebra.  I think that it would be much better if
> you could accompany your question with some short examples (just for
> avoiding misunderstandings).
> 
> What I can tell you from now is that the algebra indeed allows
> expressing restrictions on synthesis. Concerning your question about
> data encoding and classification exchange between different
> applications, I have just defined XFML+ (an extension of XFML) which
> allows representing and exchanging algebraic expressions as well.
> Some information can be found in:
> http://www.csi.forth.gr/~tzitzik/XFMLplus/
> 
> In addition, I would like to inform you that my Web page now 
contains
> 2 articles about the Compound Terms Composition Algebra:
> (a) the one presented in EJC'2003 which gives all that one might 
want
> for understanding the algebra,  and
> (b) the one to be presented in ODBASE'2003 which describes the 
formal
> semantics and explains why we cannot use the standard logic-based
> formalisms (like Description Logics) for obtaining what we can  with
> the Compound Terms Composition Algebra.
> 
> Regards,
> 
> Yannis
> 
> --- In [email protected], "Aida Slavic"
> <aida@a...> wrote:
> > Yannis,
> >
> > I need a bit of help in understanding the scope of 'compound terms
> > composition algebra'.
> >
> > In managing facetted classification schemes there are three kind 
of
> rules
> > that needs to be encoded
> > properly (apropriate data binding).
> >
> > -facet citation order rule (the order of facets in compound term 
to
> achieve
> > horizontal order with
> > decending concretness)
> > -filing order rules (the order compounds terms when they are filed
> to
> > achieve the hierarchy in
> > knowledge display: general to specific)
> > -restriction on synthesis and rules on combination between facets
> > (the areas of application of certain facets)
> >
> > (NB I am not talking here about encoding semantic relationships
> that are
> > dealt with on the level of facet
> > itself)
> >
> > While the first two rules are traditionally solved with
> appropriate 'facet
> > indicators' in classification notation,
> > followed by appropriate data coding of this notation, the last one
> needs a
> > special attention.
> > This is something that is usually  'handed over' to the 
application
> or tool
> > itself
> > (program memory) and I have never seen it coded on the level of
> > classification data.
> >
> > Am I right to think that 'compound terms composition algebra' 
takes
> care
> > about all of these rules?
> > Does it provide solutions for data encoding as well (e.g. for the
> the
> > purpose of classification
> > exchange and sharing between different applications).
> >
> >
> > Aida
> >
> 
> 
> 
> Yahoo! Groups Sponsor
> 
> 
> 
> ----------
> Thanks for playing. To unsubscribe from this group, send an email 
to:
> [email protected]
> 
> 
> 
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.


------------------------ Yahoo! Groups Sponsor ---------------------~-->
Buy Ink Cartridges or Refill Kits for Your HP, Epson, Canon or Lexmark
Printer at Myinks.com. Free s/h on orders $50 or more to the US & Canada. http://www.c1tracking.com/l.asp?cid=5511
http://us.click.yahoo.com/l.m7sD/LIdGAA/qnsNAA/0bmwlB/TM
---------------------------------------------------------------------~->

----------
Thanks for playing. To unsubscribe from this group, send an email to:
[email protected]

 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/