On The Information Mining Algorithms Of Networked Collective Intelligence Based On Causality

Many advances have been achieved in the study of collective behavior of animal groups and human beings.However,the coordinated behavior of animals is different from that of human beings.Generally,temporal networks describing animal coordinated behavior have higher orders than those of human beings,which are necessary to be described with high-order models.In this study,we propose a method using the time series data of collective behavior to determine the optimal maximum Markov order of animal groups so as to reflect the maximum memory capacity of the interacting network.The method combines a time-delayed causal inference algorithm and a multi-order graphical model.On the one hand,based on the causation entropy of information theory,causation inference algorithm constructs the network through aggregative discovery of causal nodes and progressive removal of undirected causal nodes.On the other hand,the multi-order model of Markov chains detects the optimal maximal order of sequential data extracted from collective behavior networks by statistical test on likelihoods of candidate models.In this paper,we apply the method to the data of pigeon flocks,dogs and a group of midges to build their temporal casual networks through the time-delayed causal inference algorithm.After counting the information flow sets,we determine their optimal maximum order and construct high-order De Bruijn graphs to describing their interacting relationships.The optimal maximum orders of six human behavior datasets are also investigated to make comparison with animal coordinated behaviors.The optimal maximum order of the six human communication datasets indicate a first order result which is smaller than all the optimal maximum orders of the pigeon flock,the midge group and the dog group.We can infer from the result that in the communication and cooperation of human society,the length of the Markov chain of the information flow is obviously smaller compared with that of animal groups,which implies a high degree of coordination and information sharing between animal individuals.On the other hand,humans can also accomplish very complex cooperation without frequent exchange of information,which reveals that the amount of information contained in a single human communication is much more massive than that of an animal individual.In many experiments,the coordinated behavior of animals tends to have a higher order than human communication.This is an empirical conclusion,but it is satisfied in most experiments.Most temporal network data of animal movements can be effectively analyzed by our method,which may provides a practical and promising solution to detect the optimal maximum order of collective behavior.