Reconstruction of opinion dynamics network with bounded confidence via compressive sensing

In recent years opinion dynamics has been widely used in political and economy. In this article we present a opinion dynamics model where agents show discrete actions and continuous opinions in a social network. Our model combines bounded confidence and real society network which differs from previous regular networks. In real life, the structure of social networks are often unknown, so uncovering the interacting structure of the underlying network is the key to get final opinions. Based on the sparse of social networks and observed time series, we use compressive sensing, an efficient way to reconstructing the social network where the opinion updates take place. We find that with a smaller threshold a better success rate of recovering is obtained. Also with the threshold increasing more final opinions survive at last. And the density of network also affects the final opinions.