Language trend prediction based on adaptive composite language network

The evolution of global language distribution is a complex dynamic evolution system which is very difficult to analyze quantitatively. In order to solve this problem, an adaptive composite language network (ACLN) in which the population was chosen as the language carrier and the space region as the boundary, is presented. The ACLN network consists of a main network based on the nearest neighbor principle and several fully-connected language subnets. The bridge between the main network and each sub-network is established with the language family. Probability transfer matrix is adopted to describe the evolution process of ACLN network and the (Particle Swarm Optimization) PSO algorithm is applied to optimize probability transfer matrix adaptively. The model error analysis shows that the ACLN network is credible and the evolution of language distribution can be effectively quantified.