Provenance Identification in Diffusion Networks with Incomplete Cascades
作者单位：College of EngineeringPeking University School of Information Technology & ManagementUniversity of International Business & Economics
会议名称：《第37届中国控制会议》
会议日期：2018年
学科分类：07[理学] 070104[理学-应用数学] 0701[理学-数学]
基 金：supported by NSFC(No.61673027) National Basic Research Program of China(973 Program,No.2012CB821200)
关 键 词：Provenance Identification Diffusion Network Susceptible-infected(SI) Model Maximum Likelihood(ML) Estimator Pruning Rule
摘 要：We consider the problem of provenance identification for diffusion processes on social networks based on cascade information observed from a small set of nodes during the observation windows. The diffusion dynamics are described by the susceptible-infected(SI) model and a constrained maximum likelihood(ML) estimator is formulated to maximize the probability of the diffusion provenance. The identification approach consists of two steps: first, a pruning rule is defined to obtain a set of suspected provenance nodes from susceptible node observers;and then a correlation coefficient is maximized to find the provenance node from the candidate ones. Experiments on synthetic networks and real-world networks verify the effectiveness of our approach.