Ontology (computer science)

from Wikipedia, the free encyclopedia

Ontologies in computer science are mostly linguistically formulated and formally ordered representations of a set of concepts and the relationships between them in a certain subject area. They are used to exchange “ knowledge ” in digitized and formal form between application programs and services. Knowledge encompasses both general knowledge and knowledge of very specific subject areas and processes.

Ontologies contain inference and integrity rules, that is, rules for drawing conclusions and ensuring their validity. Ontologies have experienced an upswing in recent years with the idea of ​​the semantic web and are therefore part of the knowledge representation in the sub-area of artificial intelligence . In contrast to a taxonomy that only a hierarchical sub- outline form, provides an ontology network is information with logical relations.

Publications usually speak of an “explicit formal specification of a conceptualization” (concept formation). Since ontologies have a high semantic expressiveness, they are also suitable for representing complex data models or knowledge representations. This means that the consensus of a large number of partners can be formalized with the help of an ontology, even in collaborative projects.

purpose

Ontologies serve as a means of structuring and for data exchange in order to

  • to merge already existing knowledge
  • to search in existing knowledge stocks and to edit them
  • Generate new instances from types of knowledge .

Most of the known applications do not know any individual instances and are limited to scientific purposes for systematizing the use of conceptual spaces. Ontologies are known for genetic data in bioinformatics or spatial information in geosemantics .

New applications are to be expected if the ontologies are used as types for the instantiation of individual information concepts, for example in human medicine for case-specific medical documentation, the patient file. Applications that have already been developed in human medicine have not yet established a connection between known classification systems in clinical practice. Instead, they have so far only been linked to individual classifications for scientific work.

Experiments on the profitable use of ontologies in business application software have been published by SAP .

The strength of ontological concepts lies in the bridging function between different classifications and neighboring conceptual worlds: They allow the conceptual work to be detached from fixed text templates and text modules and the transition to changing compilations of semi-finished texts for drafting individual texts.

construction

Similar to a database , in which the structure ( database schema ) and content ( data ) form a whole, the rules and terms also belong together in an ontology. While classic databases have no information about the meaning of the stored data, ontologies have a formal description of the data and rules about their relationship. These rules make it possible to draw conclusions from the existing data, to recognize contradictions in the data and to supplement missing knowledge independently from the existing. These conclusions are derived through inference , i.e. through logical inference .

Under "ontology learning" (perhaps to be translated as "ontological learning") the process can be described in which an ontology acquires additional knowledge through automatic procedures and thereby increases in scope and structure. Inferences play an important role for this. In this process, knowledge is generated by an automated process, while ontologies otherwise gain knowledge through input from human experts.

The possibility of relations via relations (referred to as reification in RDF ) and rules is relatively seldom used in practice due to their complexity, among other things, although it is precisely these characteristics that distinguish ontologies from other conceptual systems.

Components

(Examples regarding the ontology "map")

  • Terms: (in English: concepts , often translated with the wrong friend “concepts”): The description of common properties is defined as a term (eg “city” or “country”). Terms are also known as classes. These can be arranged in a class structure with superclasses and subclasses.
  • Types: Types represent object types in the ontology and represent the available types in classes. These are generated using previously defined terms and referred to as types (e.g. city as the type of the term topological element of the class points or flow as the type of Conceptual topological element of the class lines )
  • Instances: Instances represent objects in the ontology. They are generated on the basis of previously defined terms and are also referred to as individuals (e.g. Munich as an instance of the term topological location of the city type or Germany as an instance of the term topological location of the country type).
  • Relations: Relations are used to describe the relationships between the instances (e.g. the city of Munich is in the country of Germany) and are also referred to as properties.
  • Inheritance: It is possible to inherit relations and properties of the terms. All properties are passed on to the inheriting element. Multiple inheritance of terms is generally possible. By using transitivity, instances can be set up in a bottom-up hierarchy. This is called delegation.
  • Axioms: Axioms are statements within the ontology that are always true. These are usually used to represent knowledge that cannot be derived from other terms (e.g., "There is no train connection between America and Europe").

Ontology types

Basically ontologies are divided into two types:

  • Lightweight ontologies contain terms, taxonomies, and relationships between terms and properties that describe them.
  • Heavyweight ontologies are an extension of lightweight ontologies and add axioms and restrictions to them, making the intended meaning of individual statements within the ontology clearer.

Creation

An ontology depends on who is using it. For example, in the case of an ontology about wines for a restaurant, it can be important to include dishes that match the wines in the ontology. If, on the other hand, the user is a wine bottler, the food area should be completely uninteresting. On the other hand, it is important for the bottler which different types of cork and bottle exist.

Various formalized process flows have been proposed for the creation and expansion of ontologies. The methods according to Holsapple and Joshi, according to Gómez-Pérez or Uschold are increasingly dedicated to the cooperation of experts in the field of ontology and computer scientists or general formalists. Automatically supporting methods either have the goal of carrying out a complete construction of the ontology (such as the method by Alexander Mädche ) or to expand existing ontologies by suggesting terms (e.g. the method by Faatz and Steinmetz). When creating ontologies, the merging of existing ontologies can also be of interest. There is a formal procedure for this according to mute and girl. The “Ontoverse” project pursues the approach of building an ontology collaboratively and implementing it as a wiki.

Example ontology

Example ontology

The illustration opposite shows the functional principle of an ontology. The upper level shows the ontology, which contains concepts and relations. Concepts are represented by ellipses and relations by arrows. The rectangles represent simple containers for information. The relations connect two terms with each other and at the same time limit them, for example a work of art is created by an artist.

Terms can be used for inheritance. For this reason, painters and sculptors also have the surname and first name relations. The thick arrow indicates inheritance. The two relations strikes and paints and gemaltVon and geschlagenVon are inherited relations of produced and manufactured . The original relation properties are retained, but can be expanded.

The relations paints and paintedVon have inverse relationships to each other, whereby further logic is integrated into the ontology, which enables a painter to draw conclusions about his works of art and, conversely, a picture about his painter.

The lower level of the figure shows instances of the ontology. These are represented by a black point. The abbreviation (I1) stands for the unique resource name of the instance. In the Semantic Web, a URI is used for identification. The instance of the painter Raffaello Santi has a special feature. This uses existing instances, namely I3 of the oil drawing type and I6 of the Galleria dell'Accademia type.

Ontology editors

Various software tools support the construction of ontologies in various ontology languages.

Ontology languages

Formal languages ​​for describing ontologies include the RDF schema , DAML + OIL , F-Logic , the Web Ontology Language (OWL) propagated by the World Wide Web Consortium for the semantic web , the Web Service Modeling Language (WSML) and the Topic Maps standardized under ISO / IEC 13250: 2000 . The Knowledge Interchange Format (KIF) is also used occasionally.

history

Originally, ontology as the doctrine of beings is a philosophical discipline and part of metaphysics .

Charles S. Peirce and Edmund Husserl are to be mentioned as a forerunner of an explicit formalization of the ontology concept . Alonzo Church 1958 and Willard Van Orman Quine also had a formal view of the philosophical ontology . Quine put forward a concept of ontology that broke with the tradition of the classical conception of the concept of ontology in philosophy. According to Quine, “to be” means: to be the value of a bound variable. In On the Way to Truth there is the thesis: "Empirically of importance in an ontology are only the said neutral nodes that it contributes to the structure of the theory."

In the field of artificial intelligence, the term “ontology” was introduced in the early 1990s by an article by Neches et al. and subsequent publications popular.

From then on, the term “ontology” spread as an explicit formalization, was used in artificial intelligence research and taken up by bioinformatics and other subjects.

In 1999, Tim Berners-Lee presented his vision of the Semantic Web in the book Weaving the Web . The article The Semantic Web by Berners-Lee et al. Is cited many times in this context . a. from 2001, in which he also describes the use of ontologies in connection with the Semantic Web .

See also

  • Formal concept analysis . Ontologies in the sense of computer science can be represented mathematically using the means of formal concept analysis. So there is a close relationship between the two areas.
  • Systems theory . While ontology focuses on capturing basic structures or recognizing and deriving these structures in large amounts of data, systems theory tries, at least in the technical area, to capture more far-reaching aspects of such structures, e.g. B. quantitative aspects and their behavior over time.

literature

Understanding ontology

Biomedical ontology

Applications

Web links

Wiktionary: ontology  - explanations of meanings, word origins, synonyms, translations
Commons : Ontology (computer science)  - collection of images, videos and audio files

Individual evidence

  1. ^ A b T. R. Gruber: A translation approach to portable ontologies . In: Knowledge Acquisition . tape 5 , no. 2 . Academic Press, 1993, pp. 199–220 ( ksl-web.stanford.edu [accessed February 22, 2017]). ksl-web.stanford.edu ( Memento of the original from February 10, 2007 in the Internet Archive ) Info: The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice.  @1@ 2Template: Webachiv / IABot / ksl-web.stanford.edu
  2. ↑ Christina Feilmayr, Wolfram Wöß: An analysis of ontologies and their success factors for application to business . In: Data & Knowledge Engineering . 2016, pp. 1–23. Retrieved May 23, 2017.
  3. ^ Daniel Oberle: How ontologies benefit enterprise applications . Ed .: Semantic Web journal. tape 5 , no. 6 . IOS Press, 2014, doi : 10.3233 / SW-130114 ( semantic-web-journal.net [PDF; accessed on February 22, 2017]).
  4. ↑ wwwalt.phil-fak.uni-duesseldorf.de (PDF)
  5. ^ Ontological Commitment . In: The Journal of Philosophy , 55, pp. 1008-1014
  6. ↑ Relevant texts are From a logical standpoint , engl. Orig. 1961 and Ontological Relativity , engl. Orig. 1969
  7. ↑ (WVO Quine: On the way to truth , §13 Resolution der Ontologie, Paderborn et al. 1995, p. 45.). See also substitute function
  8. ↑ Robert Neches, Richard Fikes, Tim Finin, Thomas Gruber, Ramesh Patil, Ted Senator, William R. Swartout: Enabling technology for knowledge sharing . In: AI Magazine, Volume 12, Number 3, 1991 isi.edu
  9. ^ M Ashburner , CA Ball, JA Blake, D Botstein, H Butler, JM Cherry, AP Davis, K Dolinski, SS Dwight, JT Eppig, MA Harris, DP Hill, L Issel-Tarver, A Kasarskis, S Lewis, JC Matese , JE Richardson, M Ringwald, GM Rubin, G Sherlock: Gene ontology: tool for the unification of biology. The Gene Ontology Consortium . In: Nat Genet. , 2000 May, 25 (1), pp. 25-29, PMID 10802651
  10. ↑ Tim Berners-Lee, Fischetti, Mark Fischetti: Weaving the Web . Ed .: HarperSanFrancisco . 1999, ISBN 978-0-06-251587-2 , pp. chapter 12 .
  11. ↑ Tim Berners-Lee, James Hendler, Ora Lassila: The Semantic Web: a new form of Web content that is meaningful to computers will unleash a revolution of new possibilities. In: Scientific American , 284 (5), pp. 34–43, May 2001 (German: My computer understands me . In: Spectrum of Science , August 2001, pp. 42–49)