Knowledge Graph is a graph of data that represents knowledge of the real world, with nodes standing for entities and edges standing for the relationships between them.
The phrase has appeared in the literature since at least 1972, according to the 2021 ACM Computing Surveys paper "Knowledge Graphs" by Aidan Hogan and co-authors, which traces its modern usage to 📝Google's 2012 announcement of its Knowledge Graph. That product, announced by Amit Singhal under the title "Introducing the Knowledge Graph: things, not strings," is one instance of the general concept; at launch it held more than 500 million objects and 3.5 billion facts and relationships.
The same structure underlies the W3C's 📝Resource Description Framework (RDF), one of the standards of the 📝Semantic Web. Each RDF statement is a triple of subject, predicate and object, and a set of triples forms a graph whose nodes are the subjects and objects and whose arcs are the predicates.
Some 📝Personal Knowledge Management tools apply the node-and-edge model to a person's notes. 📝Obsidian's Graph view draws each note as a node and each internal link as a line, and its local graph shows only the notes connected to the active note. In 📝MythOS, the 📝Graph maps a library's memos, daily entries, tags and collaborators as nodes joined by mentions, forks and shared tags.
I don't think the Semantic Web died. It went underground and became the plumbing of the modern AI stack, and the knowledge graphs at Google, Amazon and Meta are its direct descendants.