2015 Ontologies

• (Navigli, 2015) ⇒ Roberto Navigli. (2015). “Ontologies.” In: Reference Book Journal.

Subject Headings: Ontology DB, Ontology Building, Learning, Matching, Mapping, Merging.

Cited By

• http://scholar.google.com/scholar?q=%222015%22+Ontologies

Quotes

Author Keywords

• Ontologies; Knowledge representation; Lexical semantics; Semantic Web

Abstract

This chapter is about ontologies, that is, knowledge models of a domain of interest. We introduce ontologies, view them from the perspective of several fields of knowledge, and present existing ontologies and the different tasks of ontology building, learning, matching, mapping and merging. We also review interfaces for building ontologies and the knowledge representation languages used to implement them. Finally, we discuss the different ways of evaluating an ontology and the applications in which it can be used.

20.1 Introduction

In computational linguistics and computer science, an ontology is a formal representation of knowledge. Since ancient times human beings have constantly searched for new ways to express and encode their knowledge. Nevertheless, until recently this knowledge has overwhelmingly been represented by means of informal tools, such as natural language and pictures. Today, however, with the advent of computers – and the Web era – it is becoming increasingly clear that formally encoding knowledge would make possible a new generation of the Web to enable information processing at the meaning level.

In fact, ontologies are about meaning. A popular definition for an ontology is “a formal specification of a shared conceptualisation” (Gruber 1993). This definition makes it clear that we need to represent formally and explicitly our model of the knowledge we are interested in (typically, a domain) and that this model should be agreed among users, experts, communities, etc. In other words, we can say that an ontology is a set of definitions in a formal language for concepts that describe the world of interest, including the relationships that connect these concepts.

So, ontologies are about formalising knowledge. But how formal and explicit are ontologies? This question can be answered by comparing the degree of formalisation of ontologies with that of other resources such as terminologies, glossaries, thesauri and taxonomies. As can be seen in Figure 1, the degree of formalisation constantly increases from the least to the most formalised knowledge resource: unstructured text – just a string of text with no additional structure; terminology – a set of terms expressing concepts for the domain of interest (e.g. hotel, room, tourist, etc.); glossary – a terminology with textual definitions for each term (e.g. “an establishment that provides short-term lodging” as definition of hotel); thesaurus – which provides information about relationships between words, like synonyms (e.g. motel is a synonym of motor hotel) and antonyms (e.g. ugly is an antonym of beautiful); taxonomy – a hierarchical classification of concepts (e.g. a motel is-a hotel); ontology – a fully-structured knowledge model, including concepts, relations of various kinds and, possibly, rules and axioms.

20.2 Anatomy of an Ontology

20.2.1 Building blocks of an Ontology

An ontology is composed of the following building blocks:

• Concepts (or classes), which represent the basic units of an ontology. A concept identifies a meaning that the ontology creators want to include in their representation of the domain. If an ontology is lexicalised, a concept is associated with one or more terms that express it by means of language. For instance, given the car concept in the automobile domain, synonyms such as car, automobile, motorcar are typically associated with that concept. In knowledge representation languages (cf. Section 20.8), a TBox (terminological box)]] includes the set of concepts of an ontology.
• Instances (or individuals or objects), which represent the ground level of the ontology. These are objects of the real world, such as an existing car license plate number in the domain of interest (e.g. LO108ST). In knowledge representation languages (cf. Section 20.8), an ABox (assertional box) includes the set of instances of an ontology.
• Relations, which connect concepts and inviduals to one another. Among the most popular (and relevant) ontological relations we mention:

…

References

;

	Author	volume	Date Value	title	type	journal	titleUrl	doi	note	year
2015 Ontologies	Roberto Navigli			Ontologies						2015