Glossary of Terms

Definitions for commonly used terms within the semantic web ecosystem.

Term	Definition
AAA Slogan	The idea that Anyone can say Anything about Anything (AAA). This is one of the central justifications for why a framework needs to exist around the collection and collation of web based knowledge.
Asserted triple	A triple that was explicitly added to the graph by a user or via some data source - i.e. not inferred.
Axiom	This is what facts/statements/keywords etc. are called in OWL - essentially anything that says what is true within a particular domain.
Closed-World Assumption	The assumption that there is no more knowledge to find on top of what is known - e.g. if a database said someone had two brothers, you can assume that to be complete and true (i.e. there are no other brothers).
Inferencing	The process by which new knowledge is generated from existing knowledge based on rules and patterns (specifically in RDF, this means new 'triples' being added based on existing triples).
Inferred triple	A triple that was added to a graph based on some inference rule.
Namespace	A domain/authority used to group together a particular set of named resources - similar usage in computer programming. Used to avoid naming conflicts between organisations (e.g. foaf:Person vs dbo:Person).
Open-World Assumption	The assumption that there might always be new knowledge out there to discover - e.g. if a database said someone had two brothers, you cannot discount the existence of more brothers.
OWL (Web Ontology Language)	A set of classes and properties used for describing RDF data. Much richer and more expressive than RDFS.
OWL DL (Web Ontology Language Description Logic)	Limited dialect of OWL that does not implement the whole feature set - provides maximum expressiveness but with certain restrictions applied to maintain computational completeness/decidability.
OWL lite (Web Ontology Language lite)	Limited dialect of OWL that does not implement the whole feature set - primarily implements class heiracy and simple constratints only.
qname	An abbreviation for URIs that allow them to be written without having to specify the whole location (e.g. example:foo instead of http://example.com/foo).
RDF (Resource Description Framework)	A standard for representing graph knowledge as 'triples' (subject, predicate, object) or sometimes 'quads' (subject, predicate, object, graph).
RDFa (RDF in Attributes)	A standard of decorating HTML elements with RDF data (via attributes). This allows the meaning of an element to be explicitly stated and therefore understood when parsed.
RDFS (RDF Schema)	A set of classes and properties used for describing RDF data. Mainly concerned with applying set theory and defining heirachies to RDF knowledge.
RDFS Plus	A set of standards that includes the whole of RDFS as well as a small subset of the most commonly used OWL standards.
Reification	A pattern for modelling a statement about an existing statement.
Relationship Transfer	Making an inference that if something is related to 'A' by property 'p' then it should also be related to 'B' by property 'q' - e.g. 'If something is a living thing then it must be something that respires'.
SHACL (Shapes Constraint Language)	A W3C recommendation to handle use cases around validation and interface definition.
Skolemisation	The act of replacing all blank nodes in a graph with freshly minted IRIs (i.e. real nodes).
SKOS (Simple Knowledge Organization System)	A standard set of classes and properties for representing any kind of controlled vocabulary (e.g. a thesaurus or a syntax). Endorsed by W3C.
SPARQL (Sparql Protocol And Rdf Query Language)	Query language for asking questions of RDF-compatible data. Roughly synonymous with other query languages such as SQL and Cypher.
SWRL	Semantic Web Rule Language - proposal for a sublanguage containing some, but not all, of OWL.
Triple	The main data structure within RDF. A triple consists of a subject, a predicated (or property) and an object, for example 'bob (s) knows (p) dave (o)'
Turtle (ttl)	A serialization syntax for RDF (one of the most popular and human-readable formats).
Unique Name Assumption	The assumption that two unique IRIS must refer to two unique entities - this is not the case for owl reasoning - if a database contains two records representing what look like two different entities (e.g. two people) the an OWL reasoner cannot discount the possibility that they are in fact both the same person unless the data explicitly says so.
URI (Uniform Resource Identifier)	A unique set of characters that identify something on the web. Similar to URLs but more general, they permit more characters and can represent more than just web pages (e.g. can be used to represent even real-world objects).