An Interactive Relationship Network Based Collective Intelligence Calculation Method

Quantitative methods of collective intelligence, as a means of describing the degree of intelligence emergence in the process of gaming or collaborating, are of great significance for the analysis and optimization research of swarm intelligence. The interactions between heterogeneous intelligent individuals stimulate the emergence of collective intelligence. However, current means of quantifying collective intelligence fail to take the behavioral interactions between individuals into account. In this paper, we provide a method for calculating collective intelligence based on the network of interactive relationships. This method distinguishes the interactive relationships among individuals as cooperation, defection, and inaction. Under each of these three relationships, the individual's contribution to collective intelligence is calculated differently. Firstly, the interactive relationship network was divided into three sub-networks containing only a single interactive relationship according to the three relationships used. Then the collective intelligence in each of the three sub-networks is calculated. The total collective intelligence of the interactive relationship network is finally obtained as the sum of the collective intelligence of the three sub-networks. We designed a series of simulation experiments of the intelligent individual game to demonstrate the effectiveness of our proposed approach. In the game, intelligent individuals compete for the right to participate in a job, where the cost of the job is shared by the participating individuals and the benefits are shared by the whole group. The effects of tax rates, learning rates, benefit-cost ratios, and network averaging degrees on collective intelligence are investigated by simulations respectively. Our method provides a potential approach to explore collective intelligence emergence phenomenon.