Ego-centric Network Sampling in Viral Marketing Applications

Marketing is most successful when people spread the message within their social network. The Internet can serve as an approximation of the spread of messages, particularly marketing campaigns, to both measure marketing effectiveness and provide data for influencing future efforts. However, to measure the network of Web sites spreading the marketing message potentially requires a massive amount of data collection over a long period of time. Additionally, collecting data from the Internet is very noisy and can create a false sense of precision. Therefore, we propose to use ego-centric network sampling to both reduce the amount of data required to collect as well as handle the inherent uncertainty of the data collected. In this paper, we study whether the proposed ego-centric network sampling accurately captures the network structure. We use the Stanford-Berkeley network to show that the approach can capture the underlying structure with a minimal amount of data.