Knowledge graphs are a dying technology
A knowledge graph, also known as a semantic network, represents a network of real-world entities—i.e. objects, events, situations, or concepts—and illustrates the relationship between them. This information is usually stored in a graph database and visualized as a graph structure, prompting the term knowledge “graph.”
In practice knowledge graphs are snapshots of a set of objects and some factual connections between them:
Seemingly impressive, this example highlights the limitations of knowledge graphs. Firstly, some of the graph is trivial. Secondly, some of the labels are probably wrong (surely Global Investment Inc controls Big Bucks Cafe?), meaning that the inferred facts are wrong too. Thirdly and relatedly, what does the metadata controls mean anyway? Direct control? Franchise?
Get Ben Toth’s stories in your inbox
Join Medium for free to get updates from this writer.
Less visible are limitations such as the inevitable incompleteness of the graph. And a time issue — if Global Investment parts way with Big Bucks Cafe the graph needs reworking.
Adding metadata in 2023 is like adding an engine to a horse drawn cart and saying ‘it goes a bit faster this way’. Using foundation models to create a knowledge graph is the same thing.
