“A Framework for Representing Knowledge” was written by Marvin Minsky and issued as MIT Artificial Intelligence Laboratory Memo 306 in June 1974; the source cited here is the original memo in MIT’s DSpace archive. A condensed version was reprinted in “The Psychology of Computer Vision” in 1975 and again in “Mind Design,” and it became one of the most cited works in knowledge representation.
Minsky’s central proposal is the frame: a data structure that captures a stereotyped situation, such as being in a particular kind of room or attending a child’s birthday party. A frame comes with slots to be filled in, default values for what one normally expects, and attached procedures for what to do when expectations are not met. When you enter a new situation, Minsky argued, you do not reason from scratch; you retrieve a remembered frame and adapt it, changing details to fit reality. Frames are linked together into networks so that recognizing one situation activates related ones.
The paper was partly a reaction against approaches that tried to build intelligence purely from logic and small isolated facts. Minsky argued that human-like understanding depends on large, chunked, expectation-laden structures of knowledge. Frames influenced a generation of knowledge-representation work, fed directly into object-oriented and frame-based languages, and shaped how expert systems organized what they knew.
Why business readers should care: the idea that knowledge is best stored as structured templates with defaults and exceptions is the conceptual ancestor of schemas, object models, and the ontologies that organize enterprise data and the knowledge graphs behind modern search and assistants.