Networks stand at the center of our society, including social networks in virtual online space and technology networks among communication devices. Making socio-technological networks robust is becoming a paramount concern for national security, disaster relief, and economic stability. This project brings together a truly inter-disciplinary team to develop the fundamental research methodologies and perform large-scale human subject experimentations towards this goal.
Intellectual Merit: (1) Developing Foundational Tools for Robust Networking. We draw from a suite of mathematical and statistical tools on two driving applications: (i) robustness against shocks, from the angles of social structures, policy influence, and communication recovery, and (ii) topology's impact on information value and propagation. (2) Interacting Across Disciplinary Boundaries. This team consists of four faculty members from three departments: Electrical Engineering, Sociology, and Political Science. Collectively the researchers draw upon expertise ranging from large-scale, social-network-based human behavior study to stochastic optimization over heterogeneous communication devices. (3) Bridging Theory-Practice Gap. A major bottleneck to social network study is the lack of an experimental testbed involving both innovative research agenda and human subjects. There are two sets of unique experimental platforms developed by the team: Online Gaming Community and Sharing Mart.
Broader Impacts. In addition to innovations in curriculum development across three departments, this team also actively reaches out to a variety of communities, from Non-Government-Organizations to high school students and online gamers, from major companies in the communication networking sector to undergraduates interested in the intersection between social sciences and engineering.
In this inter-disciplinary team, four faculty come from three different departments, including engineering, sociology, and politics, and together study the parallels and interactions between social and technological networks, through both theoretical modeling and data-driven experimentation. In addition to research activities and publications, the team also carried out substantial amount of education activities and curriculum development, including the creation of a new undergraduate course, its open online version, and the corresponding textbooks, as well as active outreach to industry and government collaborators. The research outcome of the project include the following: apply network sampling theory to populations at risk of AIDS/HIV, take massive amount of mobile phone usage data in Afghanistan to study social fabric there, take massive amount of Twitter data to quantify biases in movie ratings and election coverage, unify various quantifications of fairness in social and technological networks, explain the robustness presence of short paths in social networks, quantify robustness of communities in social and technological networks. For example, the team has carried out several with high visibility and substantial novelty, including analysis of cell phone tower construction's correlation with insurgent activities in Afghanistan and Iraq, modeling of spread of HIV in human networks in collaboration with global health authorities, a unifying, axiomatic construction of fairness in resource allocation in both human and tech networks, large-scale analysis of tweets during Oscar and presidential election (covered extensively by online media), and quantification of robustness of communities in Yelp (winner of the 2014 Grand Prize of Yelp Data Challenge). We have shown that: Respondent-driven sampling is less accurate than previously believed, and respondent-driven sampling confidence intervals are misleadingly narrow. The expansion of the cellular communications network in Iraq from 2004-2009 appears led to increased levels of insurgent violence in areas that received coverage for the first time. There is no evidence that extending cellular coverage to an area causes convergence in mean attitudes across populations in Afghanistan. There is modest evidence that introducing cell phone coverage leads to a convergence in the distribution of attitudes within ethnic groups in Afghanistan. Incentive-based strategies work better than network-based strategies for limiting aggregate social behaviors such as insurgency or financial panics. Hubs have limited effect in reducing the average message delivery time in small world networks. We also developed respondent-driven sampling (RDS), a method for estimating the risk behavior and disease prevalence in hard-to-sample groups. We have demonstrated that the effect of the media on aggregate behavior is more complex than simple generalization from the individual would imply. Social network interactions can amplify media bias, leading to large swings in aggregate behavior made more severe when individuals can select into media outlets. Twitter users are in general more positive about new movies than users of IMDb and Rotten Tomatoes, but the same cannot be said for Oscar-nominated movies. Finally, we found that hype-approval factors, IMDb ratings and box-office outcomes are not well-correlated for the movies studied. Quantifying political bias by analyzing large volume of tweet and retweet data can generate a useful scale of relative bias across social media outlets.
