Honeybee-inspired algorithm optimises destruction of social networks
Inspired by the collective behaviour of a honeybee colony, a novel algorithm – an “artificial bee colony” has been shown to be effective at dismantling social networks, such as those linked to organised crime.
Whether a ring of mobsters based in a city, or an online network of jihadi terrorists, law enforcement tends to use traditional methods to dismantle dangerous social networks.
The classic approach to defeating these networks is to identify their major players, and focus attentions on warning, detaining or even killing them. However, this method does not ensure that the resulting network is sapped of its organisational and reconstructive power.
“In order to achieve the most effective way of dismantling a network it is necessary to develop and put into action an optimisation process that analyses a multitude of situations and selects the best option in the shortest time possible,” says Professor Humberto Trujillo Mendoza, a psychologist based at the University of Granada and one of the researchers behind the project.
“It’s something similar to what a chess program does when identifying, predicting and checking the possible steps or paths that may occur in a game of chess from a given moment and movement.”
Professor Trujillo Mendoza and his colleagues turned to honeybee colonies for inspiration, using what they call a “swarm intelligence approach” inspired by their foraging behaviour. Colonies give rise to collective behaviour that allows members to exchange information and react efficiently to threats based on their roles.
“Bees form fairly well organised societies, in which each member has a specific role,” said Dr Manuel Lozano Márquez, also of the University of Granada. “There are three main types: scout bees, which are looking for food sources; worker bees, who collect food; and supervisor bees, who wait in the colony.”
The information exchange and communication between the different types of bees was simulated with the algorithm, rendering it an “artificial bee colony”.
The algorithm detects the most dangerous actors within a social network and processes the multitude of relationships between them. By attacking the actors most vital to the organisation of the network first, the algorithm can optimise its destruction. Results reported in Information Sciences suggest that this approach improves on the classic strategy for attacking social networks.
The authors suggest this new approach could help law enforcement authorities make decisions and act more efficiently in combating dangerous social networks.