[Person-ontology] Methodology using an upper ontology

[Philippe Martin]

> > Why not use an existing upper ontology?
>
> ... Why use it? What would it be useful for?
> ... BOF requires us to classify ocean wave 'properly'.
> In particular, we must decide whether or not it is
> a continuant. Well, is it or isn't it? Do you know?
> I don't. Moreover, I don't CARE. ...
> ... the top-down track is usually treated as authoritative,
> since to change an entrenched 'upper' ontology is seen as
> harmful to interoperation. 

Indeed. My view is that a general-purpose ontology should 
- also include and organize general "word meanings" (such as
  those of WordNet for example); it cannot categorize these
  general "word meanings" under very precise distinctions;
- include and organize the distinctions proposed by many 
  top-level ontologies (the more the better; no, this does not
  lead to inconsistencies if this is propery done), and when 
  needed, specialize the general "word meanings" according to 
  those distinctions;
- NOT be a static file but a dynamic one, that is, an 
  ontology that can be completed by people (according to
  some protocols) in order to support knowledge sharing 
  among those people (otherwise they independently create
  extensions and these extensions are then not only difficult
  to interpret, compare and merge but they have to be merged
  again and again by people).
This is what I did with WebKB-2 and its MSO (multi-source ontology).


> My understanding was that the person ontology was intended 
> to be able to represent all information about a person that
> might need to be publicly available or available to other 
> agencies (such as medical information). No?

> I think it's even greater to clarify the rationale of what
> we are encoding.

Indeed. 
1) If the knowledge modeling that will be permitted by this
Person ontology is for general knowledge organization (and then
retrieval) purpose, the Person ontolog must (also) be a large
general ontology essentially organized by relations such as
subtypeOf, instanceOf, identicalTo and partOf (physicalPartOf,
subProcessOf, etc.).
2) If the first case is true but more precise applications are also 
expected, the general "word meanings" should be specialized by more
precise categories to which more precise definitions/axioms/rules 
should be associated. Again, it is important that the end users of 
the Person ontology are allowed to do these additions. If category
identifier prefixing, editing protocols and knowledge filtering mechanisms 
are proposed (as in WebKB-2, for example, although "spam categories"
issues are not yet well handled at all), there is NO harm in 
letting them add categories.
In these two particular cases, there is a lot of work to do to create
a "good" large ontology, but apart from defining what "good" means
(e.g., what lexical and structural convenions are to be followed)
are there many other questions to debate about?

Let's also not forget the suggestion of Aldo:
> Tip for this list: ontology design is now a bit more focused 
> than 5 years ago: we need to clearly state the tasks (e.g. in 
> the form of "competency questions") that motivate the creation
> of a person ontology, and then verify what modeling solutions 
> (e.g. in the form of patterns) can help satisfying that task. 


Philippe