Search Mailing List Archives
[protege-discussion] Doubt about inheritance
mdebellis at pacbell.net
Wed Apr 17 04:03:38 PDT 2013
I've only been working with OWL a few months but I've done a lot of OO modelling. On the surface they seem similar but I think they are very different. For example, I think it would be a mistake to ever think that you could equally use Protege or UML to model a problem. (Where as for example different OO Methodologies I think apply pretty equally across domains). I think there isn't much overlap between the problem space where OO would be a good solution and the problem space where OWL would be. In my rather brief experience so far the main differences are:
1) Open World (OWL) vs. Close World (OO) assumption. This is the most obvious one and I think you could make a case that everything else flows from this.
2) Multiple Inheritance. Although some OO methodologies and tools support multiple inheritance most people that do OO (like me) never use it and think its a bad idea. (There may be some on this thread who disagree, in the academic OO world multiple inheritance is seen as legit by a lot of people but in the business world in my experience people avoid it like the plague and rightfully so IMO) In OWL type modelling multiple inheritance is essential.
3) Dealing with ambiguity. In OO ambiguity is a bad thing. If an object has multiple possible definitions its usually a sign of integrating with a legacy system and something you hope to eventually make go away. In OWL modelling the whole point is to be ABLE to deal with ambiguity because you are designing a much different kind of system that is meant to be much more flexible and open than a standard OO design.
Just my 2 cents.
--- On Tue, 4/16/13, André Luiz Tietböhl Ramos <andreltramos at gmail.com> wrote:
From: André Luiz Tietböhl Ramos <andreltramos at gmail.com>
Subject: [protege-discussion] Doubt about inheritance
To: protege-discussion at lists.stanford.edu
Date: Tuesday, April 16, 2013, 2:51 PM
Does inheritance in an ontology (superclasse/classe relationships) behave the same way as in object-oriented programming (although it is restricted to data only obviously)? That is, when inferencing is made it follows the superclass/class taxonomy as an axiomatic structure?
Thanks in advance,
André Luiz Tietböhl Ramos
-----Inline Attachment Follows-----
protege-discussion mailing list
protege-discussion at lists.stanford.edu
Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the protege-discussion