articles

Authors: Chernenkiy V., Gapanyuk Y., Kaganov Y., Revunkov G., Terekhov V. Published in № 3(69) 30 june 2017 year
Rubric: Artificial intellect

Buy an article

Download the first page

Metagraph approach for hybrid intelligent information systems description

The article is devoted to the hybrid intelligent information systems (HIIS) description using metagraph approach. The generalized structure of HIIS is proposed based on the modules of «consciousness » and «subconscious». To implement the generalized structure of HIIS it is proposed to use holonic multi-agent system (MAS). The structure of such a MAS can be described using metagraph approach. Using metagraph model it is possible to describe the structure of both static and dynamic agents to implement HIIS. Metagraph approach is one of the variants of the description of «networks with the emergence». The emergence in metagraph is achieved through the use of metavertices. It is shown that unlike the metagraph, the hypergraph is not fully «networks with the emergence». Hypernetwork fully contains the possibilities of describing the «networks with the emergence». The emergence in hypernetwork is achieved through the use of hypersimplex. Metagraphs compared to hypernetworks provides a more simple and flexible description of the «networks with the emergence». Metagraph data model, including the agents, can be presented in the predicate form, which has the property of homoiconicity. The predicate representation of data and agents allows agents of higher levels to modify the structure of the agents of the lower levels, which is an important feature of the development of HIIS. As an example of metagraph approach usage it is possible to describe the structure of a perceptron neural network. The model of neural network may be represented in metagraph form. To create, change, learn and execute the neural network different metagraph agents are used. The proposed metagraph approach may be considered as complex model for HIIS development.