A semantic network represents knowledge as a graph: nodes stand for concepts or objects, and labeled links between them stand for relationships. A node for “canary” might link to “bird” by an is-a relationship, and “bird” might link to “can fly” by a property relationship. To find out whether a canary can fly, a program follows the links upward and inherits the property from “bird.” This simple structure captures the kind of organized, associative knowledge that plain lists of facts do not.
The idea was pioneered by M. Ross Quillian, who introduced semantic networks in his work on human semantic memory at Carnegie in the mid-1960s, with his doctoral research dated to 1966. Quillian wanted to model how people store and retrieve the meanings of words. He proposed that the meaning of a word is captured by its web of connections to other words, and he modeled memory search as activity spreading outward from two concepts until their paths intersect, an idea later called spreading activation. The approach was psychologically motivated as much as it was a piece of engineering.
Semantic networks became a major branch of knowledge representation in symbolic AI, sitting alongside Minsky’s frames and logic-based methods. Researchers later sharpened the distinction between networks that define concepts and networks that assert particular facts, work that led to description logics and formal ontologies.
Why business readers should care: the semantic network is the direct ancestor of the knowledge graph, the entity-and-relationship structure that organizes web search results, product catalogs, fraud-detection links, and the factual grounding behind many AI assistants.