That Ontology Guy

What is an Ontology? The Inaugural Post by “That Ontology Guy”

(Bradley Shoebottom)

This blog posting is the first in a series describing the knowledge I have acquired as an information architect, then Knowledge Engineer (Ontologist) at Innovatia. The initial 6 parts series will cover ontology’s broadly:

1. What is an ontology? (the subject of this posting)
2. Where ontologies can be used (the whole point of the semantic web)
3. Who creates ontologies? (the knowledge engineer)
4. How to create an ontology and implement it (high level)
5. Tools used in knowledge engineering (based on previous methodology)
6. How to represent the ontology – (Querying paradigms)

After these 6 postings, I will begin to go into the details of each. Readers are invited to visit this (link) to see the how all the information fits together.

This blog will also have resources:

• links to other blogs, sites, standards,
• semantic products,
• research articles and case studies
• events

So without further ado about the purpose of this blog….

An ontology is a semantic knowledge representation of a particular subject. Think of it as the conceptual model that helps organize a person, organization, or societies thoughts about objects, people or places. The mention of “object” “people” and “places” are an ontology that describes part of nature at a high level. The ontology is semantic because it attempts to classify concepts that sometimes have multiple meanings. This definition is used by information science and those interested in knowledge management.

The name “ontology” is semantic because philosophers consider an ontology to be the study of being, existence or reality and the basic categories of being. (See Ontology in Wikipedia)  It is this last part, the categories of being, that information scientist and knowledge managers focus on. (See Ontology (Information Science) in Wikipedia) Ontologies can be powerful tools to help people find information in a complex world where there are many synonyms for a word depending on the relationship is contextualized. It is this context idea that often defines the semantic meaning of concepts. The ontology acts as a formal definition of how the concepts will be used in a particular case. Information scientists do not attempt to define the world with one encompassing ontological model, but rather they typically implement ontologies in a narrower subject area where it is easier to define concepts, their examples, and relationships.

Ontologies are often developed in a two dimensional way using forms in specialized software, but their power is the ability to represent them in 3D-like diagrams and using special software, explore the many facts of how concepts and their examples are linked together (See the Figure of a Basic Ontology)

Figure of a Basic Ontology

An ontology is made up of

• concepts,
• relationships (properties),
• examples, and
• events.

When classes and relationships are combined, they form “triples” of Subject relationship Object where the subject and object are classes. It is these triples that form the building blocks of simple to more complex queries. A concept is also known as a class. Text mining specialists call a class a T-box when they are populating the ontology with examples. (Solid blue rectangular items in above figure)  The examples are entities and sometimes called Instances or individuals of the concept or class. (Blue outlined rectangular box “International School Bus”)  Text mining specialists call these individuals “populated A-boxes” after populating the ontology. Finally, Classes and Instances are defined by their properties.

There are two kinds of properties:

• Relations or Object Properties – define how classes are related to each other, for example class Vehicle hasComponent Vehicle Component . Object properties can be defined in both directions – Vehicle Component isPartOf Vehicle.
• Attributes or Data Properties – define specific attributes of the individual Instance such as a string for its name, an integer for a measurement, or a date/time for an event instance.

Classes and properties can have axioms placed on them to prevent illogical conclusions. For example you can assert the axiom that Car is a disjunction of (or cannot be) Truck. This prevents potential inference errors in the ontology.

You can also have Rules in terms of if-then logic statements. This allows you to define relationships based on specific instance data properties for more precision in the ontology. For example, if a male has a sibling with a child, and that child is defined as a Female, then that child is classified as a Niece.

Object and Data properties can have restrictions put on them. Typical examples include maximum or minimum values (called cardinality). For example, you can restrict the Person class to Male and Female, or the Day of the week can only be in the range of Monday to Sunday.

There can also be Functions applied to data properties. For example, you can define a mathematical function to convert set of units to another when only one is specific (metric to imperial or inches to feet). (See the Quantities, Units, Dimensions and Data Types in OWL and XML website)

Lastly, you can specific events in an ontology in terms of dates and times. This allows you to track trends.

Summary of Terms in an Ontology

Ontologies can be divided into Upper Level  Ontologies and Domain Ontologies. Domain Ontologies are very specific subjects such as the domain for units, distance, or time. Upper ontologies link domain ontologies together. For example, the Events, People, and Places Ontology would be an Upper Level Ontology. A Domain ontology for Events could be one that describes weather events in more detail versus one that describes political events. Most Domain Ontologies actually import domain ontologies so that it is easier to maintain the whole ontology and keep it up to date by domain specialists.

Resources:

• OWL Pizzas: Practical Experience of Teaching OWL-DL: Common Errors & Common Patterns
• Ontology Development 101: A Guide to Creating Your First Ontology

Upcoming schedule:

1. Where ontologies can be used (the whole point of the semantic web)
2. Who creates ontologies -  the knowledge engineer
3. How to create an ontology and implement it (high level)
4. Tools used in knowledge engineering (based on previous methodology)
5. How to represent the ontology – (Querying paradigms)