As information, such as viral signals or and infectious diseases,  spreads on various kinds of real-world networks, we are interested in estimating and predicting their influences. In this presentation, we introduce a new analysis framework, based on the discrete Fokker-Planck equations, and associated computational methods to tackle this problem. It gives rise to a class of novel algorithms that work effectively for networks with unknown propagation models or parameters, or even unobserved network structure. Numerical tests are carried out on several synthetic and real-world networks to show the promising performance of the proposed method. This is based on a recent work with Shui-Nee Chow (Math, GT), Xiaojing Ye (Math, Georgia State), and Hongyuan Zha (CSE, GT)