Glossary

Assertion

In mathematics and logic theory, a phrase using a language containing formalized semantics, and whose value of true or false can be declared or deduced. In a knowledge base, assertions are axioms or facts deduced from these axioms by application of rules.

Association

Formal representation in Topic Maps of the relationship between two or more subjects. Associations permit representation of n-ary relationships, i.e. a relationship that connects several subjects with a single relationship. Associations are used in ITM for navigation and querying of a knowledge base.

Attribute

Typed data assigned to a thing, ie the value of a property of that thing. At times attibute is incorrectly used as a synonym of property.

Axiom

An assertion declared as being inherently true. The term is used in description logic to designate assertions declared in an ontology, e.g. a subsumptive relationship or a restriction on a property.

By extension, but more rare, is the use of the term to designate an assertion that is declared as true for one instance, i.e. a fact, e.g. « Peter is a Person ». Conversely, if the same assertion is deduced by a rule, it is not an axiom.

Class

Formal representation of a type in an ontology. Classes are organized in hierarchical class-subclass relationships, or subsumption axioms, which define property inheritance.

Classification

1. Application of an organizational system to a collection of resources. In particular, assignment of a documentary corpus to headings in a classification scheme or a taxonomy.

2. In its original biological meaning, a system of classes or taxons for organizing species.

3. In the document management domain, the term if often informally used for indexation, and taxonomy.

Classification rule

A rule defining the conditions under which a resource should be assigned to a specific class.
By extension, an indexation rule indicates the conditions for which a resource has a given concept for subject, and a classification rule indicates to which taxonomy heading(s) a resource should be attached.

Classification rules can be expressed in natural language and applied by human users, or expressed in formal language for use in automatic classification.

Classification scheme

A specific type of taxonomy, generally used for documentation resources having a specific subject type. For example a “Personnel classification scheme” would include standard headings for classifying documents referring to a given person, e.g. “Public Life”, “Private Life”, “Bibliography”.

Concept

Representation of a thing in a language, particularly a formal language.
In particular, an ontology element, representing an abstraction such as a class or a property, as opposed to instances representing individual objects.
A fundamental SKOS class, equivalent to a descriptor.

Concept schema

A generic SKOS term to designate any structured vocabulary capable of being described in SKOS.

Datum

A piece of information stored in a computer system
More specifically, in RDF, the value of the object of a triple other than a resource. This type of datum can have a precise physical type specification, e.g. floating point number or date.

Description

A thing’s description is the collection of information that relates to it. Usual parts of a description are the type of the thing, its properties and relationships to other things. A description is considered to be « unstructured » when it uses natural language or a multi-media representation such as video, symbol or image. A description is considered « structured » when it uses a validated standardized information format, such as a table or XML document. The description’s format often corresponds to its type.
In RDF, the description of a resource is the entire collection of triples having the resource as subject.

Description logics (DL)

A type of logic that defines formal semantics for knowledge bases and that supports reasoning. Description logic is a theoretical foundation for OWL.

Descriptor

A concept represented in a Thesaurus. A descriptor is identified by a unique preferred term in a specific language. Synonyms, alternatives, definitions and usage notes can be attached to a descriptor.

In SKOS a descriptor is represented by an instance of the Concept class.

Document

See: Documentary corpus

Documentary corpus

A collection of documentary resources relating to a knowledge base. In particular, the collection of documents that can be automatically indexed.

Documentary resource

Synonym : Document

1. A resource representing information content on a physical support, e.g. text, PDF file, web page, image, audio file, other multi-media file

2. In particular in RDF, a resource accessible over the Web via its URL

Entity

Intellectual name for a Thing used in philosophy and certain formal languages.
Named entity : a term used by text mining tools to designate individuals, and reference tables in particular.

Enumerated class

OWL synonym for Reference Table. An enumerated class defines a finite, controlled list of its instances.

Fact

An assertion concerning one or more instances in a knowledge base. A fact can be explicitly declared as an axiom, or deduced from others by rules.

Formal language

As opposed to natural language, a formal language uses a formal grammar which defines the well-formedness of expressions written in the language. A formal language may possess formal semantics.

Formal semantics

A formal language has formal semantics if it is constructed on logical bases permitting attribution of the values true or false to its assertions. RDF has formal semantics, whereas Topic Maps do not.

Formal semantics are necessary to implement reasoning tools.

Grammar

A collection of syntactic rules describing how assertions must be formulated to have meaning in a formal language.

Heading

The name of a concept used in a taxonomy.

Hierarchy

Concept organization using « parent-child » or « hypernym-hyponym » relationships. Common hierarchies found in knowledge bases are:

The most common hierarchies are organized as trees, where each concept has at most one parent. In the case where a concept has multiple parents, one speaks of multi-hierarchies.

Homonymy

A situation where different concepts have the same name. This situation prevents uses of the name as an identifier.

A vocabulary where homonymy is prohibited is a namespace.

Identification

A process permitting unique identification of a thing based on its representation or description. In knowledge bases identification relies on identifying properties or identifying rules.

Identification rule

A rule permitting to decide is two things are identical. Most identification rules use identifying properties.

Identifier

A property whose value is used to uniquely identify a resource. In the semantic web preferred identifiers are the URI.

Identifying property

A property is deemed to be identifying if its value can be used to determine if two things are unique. If a single identifying property suffices, it is referred to simply as an identifier. Usually two things are identical if several properties are equal.

For example “family name”, “first name”, “date of birth” and “place of birth” form the set of identifying properties for a person.

Index

A structured list of subjects indexing a documentary corpus, with links to elements of the corpus. In paper publications, these elements may be references to whole documents (library index), or to document sections (a single document or back of book index). An index can be a simple alphabetic list of terms, or be structured according to the vocabulary from which it is extracted, e.g. a hierarchical structure.

Indexation

The process of attaching documentary resources to subjects defined in a vocabulary. Indexation can be performed manually by documentalists, or automatically by tools applying indexation rules.

Individual

An instance of an OWL class.

Inference

Inheritance

A hierarchy supports inheritance if a characteristic defined on a concept in the hierarchy applies to all descendants of the concept.

For example:

o A property attached to a class : a vertebrate has a skeleton.

o Subsumption : A mammal is a vertebrate.

o Inheritance : A mammal has a skeleton

Instance

An individual attached to a class. For example “Peter is a Person” has the following formal translation : the individual « Peter » is an instance of the class « Person ». Attaching instances to classes constitutes population of a knowledge base.

Knowledge base

A representation, usually in a database, of a collection of facts described according to a shared ontology. Opinions differ as to whether a knowledge base includes the description of the ontology concepts, or simply the facts concerning the instances.

In ITM we use the term knowledge base for a workspace controlled by an ontology, which in turn is defined in a separate workspace. In the former workspace, the user has read-only access to the ontology, and can create instances and facts that conform to this ontology using dynamic interfaces controlled by the ontology.

The notion of knowledge base is also formally defined in description logic.

Meronymy

A hierarchical part-whole organization, for example the subdivision of a continent into countries, a company into departments, or a book into chapters and sections.

Metadata

Typed data attached to a resource, and in particular to a documentary resource. Within the context of knowledge representation, metadata is practically synonym to property.

Name

An identifying property uses in most description languages. A name is not an identifier if the languages allow homonymy.
In linguistics, a name is the most common type of signifiant.

Namespace

A vocabulary where each name is used once and has a unique meaning.
In XML, the common part of all the URIs defined in the same vocabulary.

Object

The basis element of a knowledge base.
The third part of an RDF triple, or the value of a property.

Ontology

A structured collection of classes, properties and axioms used to organize a knowledge base. For some the definition can also include the instances and facts, or just the instances of the reference tables.

OWL

A W3C standard (Febuary 2004). A member of the RDF family of languages used for ontology description. Used for the external representation of ontologies managed by Mondeca’s ITM.

Population

Creation of instances in a knowledge base. Population can be performed by human editors or via dedicated software interfaces, e.g. ITM’s web interface. Mostly population is performed in a batch database migration.

Preferred term

A word defined in a thesaurus as the unique identifier of a descriptor in a specific (spoken) language.

Property

A property is a concept allowing description of things using assertions whose subject is the thing. In particular in RDF, a property is the second element of a triple representing an assertion. The triple (France, capital, Paris) assigns the value « Paris » to the « capital » property of the subject « France ». Note that the associated value also has a property.

Published Subject Indicator

Abbreviation : PSI

In Topic Maps, a URI used to identify a subject, which provides access to a documentary resource containing an explicity description of the subject for humans. Used at Mondeca to designate a URI identifying a topic, even if this URI does not correspond to a published resource.

RDF

« Resource Description Framework » is a language initially designed for the formal description of documentary resource metadata. It has evolved to become a generic language for the representation of Web interoperable knowledge bases, and is the basis for more specific languages such as OWL and SKOS.

Reasoning

In the context of a knowledge base, reasoning tools use formal semantics to execute tasks such as coherency control, automatic classification, instance identification or fact deduction.

Reference table

A class whose instances form a controlled list.

Referent

In linguistics, the referent designates the thing represented by a sign, and in particular by a name.

Resource

A thing identified, usually by a URI, and described in a knowledge base. The R in URI and RDF.

Restriction

An axiom defining a class by a necessary or sufficient condition on a property. For example the class of all « Public Employees » can be defined by a restriction on the value of the property « employer ».

OWL supports the expression of different types of restriction.

Rule

A rule allows a system to launch a reasoning action on the content of a knowledge base. There is only a functional difference between axiom and rule. Axioms are purely static and declarative, while rules launch actions. A rule can be the dynamic expression of an axiom.

For example:

"The employer of a Person is a Compan" is an axiom connecting the classes "Person" and "Company" with the property "employer".
"If X is a Person and Y is the employer of X, then Y is a Company" expresses the same semantics in rule form

Semantic extension

In a document collection indexed with a hierarchical vocabulary such as a Thesaurus, automatic access to content indexed by narrower terms of a given term (downward semantic extension) or by broader terms (upward semantic extension).

Sign

In linguistics, the sign is the representation of a referent, the referent being the thing that it designates, and which is outside the linguistic space. A sign is composed of two parts called signified and signifier. The signifier is the visible, e.g. written manifestation of the sign. The signified is the concept or thought that occurs on seeing the signifier. This decomposition represents the « semiotic triangle ».

SKOS

RDF language designed to represent thesauri, taxonomies, and generally any type of hierarchically structured vocabulary used for classification.

Sub-class

See Subsumption

Subject

A concept used to index a documentary resource (the document’s subject).
In RDF, the first element of a triple (the subject of the statement)
In Topic Maps, the referent of a topic (the subject of conversation)

Subsumption

If a thing is a Book, then it is a document
The "Book" class is a sub-class of the "Document" class
Each instance of the "Book" class is also an instance of the "Document" class

The subsumption relation is the foundation for property inheritance. For example if a Document must have one and only one publication date, then a Book has exactly the same property.

Taxonomy

The classical meaning from the sciences, initially Biology, is the definition of a hierarchy of classes for the systematic classification of objects in a study domain, in particular living organisms. Within this context, a class as defined in this glossary is called a taxon (plural taxa), while the word “class” designates a particular level of taxon.
In information technology, taxonomy’s meaning has become vaguer. It can indicate a simple ontology, reduced to a hierarchy of classes, or it can mean a documentary classification scheme, in particular in Web interfaces. In the latter case, one usually speaks of headings and categories rather than classes.

Term

A word in a controlled vocabulary. A term usually represents a unique concept, i.e. homonymy is not permitted.
It can also means a descriptor in a thesaurus.

Terminology

A science treating the study and management of activity domain specific vocabularies.
The collection of all (systematic terminology) or some (ad-hoc terminology) specialized terms used in a particular activity domain.

Thesaurus

A controlled vocabulary used for documentary indexation, containing terms representing descriptors, and usually structured hierarchically from broader (general) to narrower (specific) meanings.

Thing

Generic term designating anything which could be represented, identified or described in an information system, and in particular a knowledge base. In this context a thing is the referent of a name or identifier (signifier) whose meaning is a formally defined concept.
Superclass of all things in the OWL language. « Everything is a Thing ».

Topic

Formal representation of a subject in the Topic Maps language family.

Topic Maps

A family of ISO standard knowledge representation languages.

Similar to RDF, the Topic Maps languages use URIs as the principal subject identifiers.

Triple

A basic element of RDF, a triple is an expression consisting of the 3 elements Subject, Predicate, Object. A triple is the assertion of a fact or axiom about the Subject. The Predicate represents the property of the subject, and the Object is the value assigned to this property. In the RDF specification, Subject and Predicate are resources, while an Object can be a resource or data.

Type

A type is a property value that is common to a number of things, and whose usual function is to distinguish these things from others. In ontology languages, the main type is usually formally represented through class membership, but it can also be represented by other properties, in particular when multiple typing is required.

For example a document can be typed according to its format, title, subject matter or functional use. The main type « Document » is indicated by the class, while the types « title », « subject matter » and « functional use » would be represented by properties.

URI

An unique identifier of a resource in the semantic Web. URI have a specific syntax, and are of two types :

URLs which permit identification and access to a resourcce, e.g. http addresses
URNs which are simply unique names associated with a specific protocol, e.g. ISBN numbers

Vocabulary

In linguistics the collection of valid words in a language
More specifically the collection of terms defined in a terminology

Workspace

ITM specific term. A workspace defines user’s access rights to a knowledge base or an ontology.

EN - Flux RSS - Mot Expression

Product documentation