semanticnetwork

Semantic networks are a form of knowledge representation used in artificial intelligence and cognitive science to organize meaning as a graph. In a semantic network, nodes represent concepts, objects, events, or propositions, while edges encode semantic relations between them. Edges may be directed and labeled with relation types such as is-a (hyponymy), part-of, instance-of, has-property, cause, or associate-with. Through inheritance along is-a links, properties can propagate from general concepts to their instances, enabling efficient inference.

Variants of semantic networks include frame-based networks and concept graphs. They were developed in the 1960s and 1970s as a way to model human semantic memory and support basic reasoning in AI systems. The idea influenced more formal knowledge representations, including ontologies, description logics, and the graph-based approaches used in the Semantic Web (RDF and OWL).

Reasoning in semantic networks often relies on graph traversal, path finding, or rule-based inference to answer queries, check consistency, or derive new facts. They are valued for their intuitive visualization and their ability to capture approximate common-sense knowledge, but face challenges in scaling, disambiguating relations, and integrating with more rigorous logics.

Today, semantic networks are frequently subsumed by knowledge graphs and ontologies, yet they remain a foundational concept in representing meaning, enabling semantic search, natural language understanding, and knowledge-based applications.

See also knowledge graph, ontology, RDF, OWL, description logic, conceptual graph.