[Top] [All Lists]

[ontolog-forum] Ontology Tools (was is-part-of: really...)

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: John F Sowa <sowa@xxxxxxxxxxx>
Date: Mon, 20 May 2013 12:00:46 -0400
Message-id: <519A48AE.3040009@xxxxxxxxxxx>
This discussion leads to issues about designing ontologies and
the tools for developing and using them.  Without appropriate
tools, it's impossible to support multiple developers and users.    (01)

To get some perspective on these issues, I recommend the
Gene Ontology (GO).  See the web page for the GO structure:    (02)

    http://www.geneontology.org/GO.ontology.structure.shtml    (03)

Closely related to the genome is the reactome, which uses the GO
plus a huge amount of information about biological processes,
biochemistry, proteins, and their relationships to genes.    (04)

Their primary notations are *diagrams*, which are mapped to both
relational databases and to several different XML-based notations.
OWL is one of them -- but not the main one.  See    (05)

    http://www.reactome.org/ReactomeGWT/entrypoint.html    (06)

You can click on the pathway image to see a sample diagram,
and you can click on a 5-minute video for a tutorial.    (07)

Some comments about the previous thread:    (08)

Matthew L
> MatthewLange || does not understand the goal(s) of || "relationships
> that are devoid of domain specific content"
> "relationships that are devoid of domain specific content" || has example
> || "plays the role of in"    (09)

That is an example of metalevel relation.  It's more relevant to the
tool developers than the people who are entering and using knowledge
about the subject.  For now, it's more distracting than enlightening.    (010)

> MatthewLange ||  likes  ||  bioinformatics    (011)

Good.  I'd like to hear your comments about the GO, Reactome, etc.,
and the tools and diagrams that are being used for them.    (012)

William F
> The goal of elimintating domain specific relationships in an ontology is
> to *nominalize or reify* all the domain-specific relations, along with
> everything else that is of substance to the ontology, so that every
> concept we want to model is on an equal footing with all the others,
> and to *separate* these domain specific relationship concepts from
> the logical underpinnings of the language.    (013)

I agree that it's important to separate the ontology from the logic.
Many notations mix them by providing special symbols and syntax for
the most frequently occurring relations.  For examples, see the
diagrams on the GO and Reactome sites.    (014)

But beyond a few very widely used categories, it's hard to find any
clear boundary between domain specific vs domain independent ontology.    (015)

I think that the people who developed the GO and Reactome tools made
reasonable choices.  Nuclear physicists and petroleum engineers have
made different choices for their tools.  Logicians and ontologists
can give them some useful advice.  But on anything related to the
subject matter, the SMEs are, by definition, the experts.    (016)

Pat C
> The use of simple nouns such as "mother" or  "sister" by themselves to
> designate a relationship presents a serious problem in determining the
> polarity: in   "X sister Y",  who is the sister - X or Y?    (017)

I agree.  In a controlled NL, you would use the syntax of the NL.
In any formal notation, you can add 'has' or 'of' as needed:    (018)

    "X has a sister Y"  <=> hasSister(X,Y) <=> isSisterOf(Y,X).    (019)

> There is an opposite consideration: if labels are easy to understand,
> users may think they understand them when in fact they don't.    (020)

The SMEs should be the ultimate arbiters.  For any domain, the SMEs
understand the expressions used by other SMEs.  See the next example.    (021)

> my translation to a non-ER dependent descriptions should have been
> there is a catalyis such that MTHFR plays the catalyzer role in that catalysis    (022)

A SME who knows the GO and Reactome ontologies would interpret MTHFR
as an acronym for MethyleneTetraHydroFolate Reductase.  Anybody who
doesn't might think it's a colloquial epithet.    (023)

John    (024)

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    (025)

<Prev in Thread] Current Thread [Next in Thread>