Semantic network (also called concept network) is a graph, where vertices represent concepts and where edges represent relations between concepts. Semantic network at the level of ontology expresses vocabulary that is helpful especially for human, but that still can be usable for machine processing. The relations between concepts that are used in semantic networks are as follows:
- synonym - concept A expresses the same thing as concept B
- antonym - concept A expresses the opposite of concept B
- meronym, holonym - part-of and has-part relation between concepts
- hyponym, hypernym - inclusion of semantic range between concepts in both directions
Semantic nets were created as an attempt to express interlingua, a common language that would be used for translation between various natural languages. A typical example is WordNet that describes relations between English words and defines the words using natural language. Parts of WordNet were translated to other languages and the links between various languages exist and can be used as the base for translation.
Topic Maps are (syntactically) standardized form of semantic networks. They allow using topics (concepts), associations (relations) between concepts (including specifying role of topic in the association), and occurrences (resources relevant to topic, in fact instances of topic). Topics, associations and occurrences are used to create ontology of a domain, and a particular topic map then uses them to expresses state of affairs in the domain.