Ontology a Practical Guide

Chapter 2

  • concerned with knowledge representation languages that allow for deduction.
    • schema are low-expressivity and low-formality, e.g. relational database, XML
    • taxonomy are in the middle, e.g. RDF/S, UML, OWL
    • Logical Theory are high, e.g. Knowledge Interchange Format and Tptp are the most expressive and most formal
  • "every software system has an ontology", it maybe just isn't made explicit
  • "semantic networks" is one of the earliest Knowledge Representation
  • Object Orientation combine procedural specification with a small amount of declarative (e.g., classes, instances, class-subclass relations), i.e., the barest minimum
  • Frame languages lack a facility for expressing rules
  • Description Logic

First Order Logic in SUO-KIF

  • no embedded formulas/propositions (only in higher-order logics)

Ontology Development Pitfalls

  • Confusing Instance and Subclass
  • Part-of vs. Subclass

Modeling Events as Relations

  • don't do (eats Bill HamSandwich);
    • presumably, model events as things unto themselves
  • "Davisonian event representation" looks a lot like what I was planning to do with hypergraphs
  • most languages that have some use in inference, like OWL, don't allow statements as arguments to relations, because it's extremely difficult to reason with.

Example

{
  "@id": "ex:Brutus-stabbed-Caesar",
  "@type": "ont.occurence.
}
  • another reason to do kebab or snake case... preservation of capitalization (if you wanna go that route)

Ontological Promiscuity / Confusing Language and Concepts

  • terms need to be well-defined

Modelling Roles as Classes

  • yay, typedb
  • can, may, should != obliged to, allowed to

Chapter 2 Exercises

An elephant is a mammal

(forall (?X) (=> (instance ?X Elephant) (instance ?X Mammal)))

Bob Likes Sue

(likes Bob Sue)

Koko is a gorilla

(instance Koko Gorilla)

Every farmer like a horse

(forall (?X) (=> (instance ?X Farmer) (exists (?Y) (and (likes ?X ?Y) (instance ?Y Horse)))))

Chapter 3: Ontologies (in the broadest sense)

  • DOLCE universals are only employed in the service of describing particulars
  • SUMO is an ontology of both particulars and universals
    • SUMO has a hierarchy of properties as well as classes

Backlinks