Knowledge Representation through Interactive Networks

Research in cognitive psychology demonstrated that human knowledge is mostly stored by the connections between neurons rather than in the nervous cells themselves. The attempt to model such an idea leads to an interactive network, where information is represented by some set of neuromimetic descriptors. These descriptors can be either objectual or attributive, i.e. specifying an object or a property. The paper aims at processing the pieces of knowledge from a certain field in order to determine the weights of the connections between descriptors. The interrogation of the knowledge base is then possible, which provides information about the characteristics of the objects and the way they are associated.