• A
  • A
  • A
  • АБB
  • АБB
  • АБB
  • А
  • А
  • А
  • А
  • А
Обычная версия сайта

Глава

Реализация программного комплекса выявления паттернов в мультиграфе на примере сети межбанковского кредитования в агентной модели банковской системы

С. 221-224.
Леонидов А. В., Васильев С. Б., Серебрянникова Е. Е., Булгакова Д. В., Коваленко А. М., Трегубов Д. О.

Graph is a universal representation of different social and economic systems. Vertices of a graph represent objects of different kind and edges describe interrelations between these objects. Vertices and edges can be characterized by a set of attributes describing different properties of objects and relations between them. A graph is called a multi-graph If its elements are characterized by multiple attributes. In such a case, different phenomena taking place in a system can be described in the form of multi-graph's subgraph, or pattern, having particular structure. The problem of recognition of the phenomena can be formalized as the problem of subgraph matching. The study presents the software system which is aimed at solution of the problem of subgraph matching in multi-graphs. The system has a multiagent implementation. One of the key structural elements of the system is a knowledge base containing computable description of patterns to be matched. The system contains a set of program agents responsible for searching for the particular pattern. As subgraph matching problem is NP complete, there is no universal efficient algorithm for matching pattern of arbitrary structure. Therefore, different searching agents may operate with different algorithms. As a result, multiagent architecture is necessary for efficient system implementation. In the study the system's work is exemplified by recognition of two different patterns in the interbank network of the agent-based model of banking system. The first pattern relates to the case of Ponzi scheme realization in the model and the second one represents the case of cyclical transition of money between three banks. The analysis of dependence between efficiency of pattern recognition and amount of noise in the data is conducted.