On 17/10/12 15:33, John F Sowa wrote:
> Dear Markus,
>
> We seem to be converging. Just a few comments:
>
>> I have not much to add. I completely agree that there is a lot left
>> to be done towards maximally useful KR approaches (for the Web and
>> otherwise).
>
> Yes. That is why I liked Tim's original proposal, which was much
> broader than the DAML final report. What is missing from current
> SW publications are the three words that Tim emphasized repeatedly
> in the proposal: diversity, heterogeneity, and interoperability. (01)
Well, I had consciously chosen the phrase "maximally useful" as opposed
to "most useful". I do not think that there will be one KR approach to
rule them all. Different approaches will be optimal for different tasks. (02)
>
>> In Oxford, we are getting some really encouraging feedback about
>> user experiences with OWL, especially in science applications. But
>> we also know (maybe better than most) that OWL is not the universal
>> solution of all KRR problems.
>
> DLs have proved to be useful for a variety of applications. The most
> fruitful combinations are hybrids: a DL for the T-box combined with
> other systems -- logic-based, procedural, or heterogeneous. (03)
Most of the applications we see are actually rather pure, in the form of
various uses of DL reasoning at design time or runtime. Of course, the
reasoner is usually called by a program, but I would not call this
hybrid KR. If you are interested, here are some recent examples
documenting the use of reasoning in biology and medicine: (04)
* David Osumi-Sutherland, Simon Reeve, Christopher J. Mungall, Fabian
Neuhaus, Alan Ruttenberg, Gregory S. X. E. Jefferis, and J. Douglas
Armstrong: A strategy for building neuroanatomy ontologies
Bioinformatics (2012) 28(9): 1262-1269
* Robert Hoehndorf, Michel Dumontier, Georgios V. Gkoutos: Identifying
aberrant pathways through integrated analysis of knowledge in
pharmacogenomics. Bioinformatics Volume 28, Issue 16. Pp. 2169-2175
* Alan Rector, Luigi Iannone: Lexically suggest, logically define:
Quality assurance of the use of qualifiers and expected results of
post-coordination in SNOMED CT. Journal of Biomedical Informatics 45
(2012) 199–209 (05)
>
> For the SMW, a much simpler T-Box would be more flexible, especially
> one that could derive a consistent multiple-inheritance hierarchy
> automatically. Have you considered FCA (Formal Concept Analysis)?
> Following is the FCA home page:
>
> http://www.upriss.org.uk/fca/fca.html
>
> As examples, look at the use of FCA for automatically deriving
> sublattices for any word in WordNet or Roget's Thesaurus:
>
> http://www.ketlab.org.uk/wordnet.html
>
> http://www.ketlab.org.uk/roget.html
>
> It's interesting to type the same word to each of them and compare
> the lattices. Their default example is 'explore'. Other examples
> are 'happy' and 'happiness'.
>
> The FCA software is available as open source. For any page in the SMW
> it could display a sublattice of ancestors, descendants, and cousins.
> (I put Uta Priss, who maintains the FCA Home Page, on the cc list.) (06)
Thanks, I am actually well aware of FCA. I have wondered about the
applications of FCA to semantic wikis for a while, but this did not
materialise so far. It seems obvious: wikis have data, FCA has methods
for visualising and analysing data. Association-rule mining could also
be interesting on wikis. For user interface design, there is a lot of
competition; lattices might have a place there, but most people are not
used to this type of display and would have to learn how to read it first. (07)
Markus (08)
--
Dr. Markus Kroetzsch
Department of Computer Science, University of Oxford
Room 306, Parks Road, OX1 3QD Oxford, United Kingdom
+44 (0)1865 283529 http://korrekt.org/ (09)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J (010)
|