Formalization of the meaning. Part 3. Formation of contexts

The previous parts of this article have attempted to begin describing an approach to building strong artificial intelligence based on sense of information. A model was proposed in which concepts were described through related points of view. The point of view was set as a context that changes the original description to its interpretation. It was shown that the meaningfulness of the interpretation can be judged by its adequacy to the memory of previous experience. The space of contexts is described, which defines a system of concepts that cover a certain subject area. In this part of the article, an algorithm is described that allows you to create an initial system of concepts based on the observable signs of phenomena, and move from it to the contexts corresponding to these concepts. For the space of contexts, a method for creating concept codes is proposed, which allows concept codes to convey the system of their internal proximity, a comparison with convolutional networks is made. Explanations of the proposed approach are considered on the example of training the visual cortex.