In a Knowledge Graph, the data is semantically linked and defined through ontologies, thereby encoding the meaning of the data alongside the data itself. Like a map of the real world, an ontology helps us understand the full picture of the domain that the data belongs to.
Once the data is semantically organized and stored, which may include a graph database, a Knowledge Graph applies a reasoner or an inference engine over the ontology, which is essentially a set of logical rules (i.e. if-then statements) that result in the creation of new knowledge, thereby expanding the Knowledge Graph continuously as new data is ingested into the graph.
Most modern Knowledge Graphs also have robust AI and machine learning systems tightly integrated into them, which dramatically expand their ability to generate new knowledge from the original data, as well as from new data that may be ingested into the graph.