Application of multi-agent coalition in network public opinion monitoring based on cultural algorithm

It would waste a lot of time and money to search agent one by one for achieving the best coalition's behalf in a cloud platform of network public opinion monitoring. In order to solve the above problem, a dynamic coalition mechanism based on the concept of ability group is proposed. The proposed mechanism can add in new agent or delete old agent from ability group according to the performance of agent. In this way, more complex tasks can be completed by selecting agents from these ability group to form agent coalition. A coalition algorithm based on the culture algorithm is proposed. The calculation process and the rules of encoding, selection, crossover and mutation are discussed in detail. The simulation result shows that the proposed approach is feasible and effective.