Yuan Yuan

Assistant Professor, Purdue University

Hi! My name is Yuan Yuan (). I am a computational social scientist and an assistant professor at the Krannert School of Management (MIS area) at Purdue University. I am also a research fellow of the MIT Connection Science and the MIT Initiative on the Digital Economy.

I am interested in

  • leveraging big data and advanced computational techniques (e.g., machine learning and causal inference) to study online social interactions and social networks.
  • developing computational techniques that combine machine learning and causal inference, with applications to online field experiments (a.k.a A/B testing).

Before Purdue, I did my PhD in Institute for Data, Systems, and Society (IDSS) at Massachusetts Institute of Technology. I received my Bachelor's degrees with honors in Computer Science and Economics from Tsinghua University.

Follow me on Twitter .

News


Selected work

[1] Yuan Yuan, Tracy Xiao Liu, Chenhao Tan, Qian Chen, Alex Pentland, and Jie Tang, "Gift Contagion in Online Groups" [Minor revision at Management Science]

[2] Yuan Yuan, Ahmad Alabdulkareem, Alex Pentland, "An Interpretable Approach for Social Network Formation Among Heterogeneous Agents", Nature Communications, 2018 [Nature] [PDF].
  • Blog: I wrote a note about this paper [Blog].
  • Abstract: Understanding the mechanisms of network formation is central in social network analysis. Network formation has been studied in many research fields with their different focuses; for example, network embedding algorithms in machine learning literature consider broad heterogeneity among agents while the social sciences emphasize the interpretability of link formation mechanisms. Here we propose a social network formation model that integrates methods in multiple disciplines and retain both heterogeneity and interpretability. We represent each agent by an “endowment vector” that encapsulates their features and use game-theoretical methods to model the utility of link formation. After applying machine learning methods, we further analyze our model by examining micro- and macro- level properties of social networks as most agent-based models do. Our work contributes to the literature on network formation by combining the methods in game theory, agent-based modeling, machine learning, and computational sociology.
  • Datasets: We provide 6 full (un-sampled) social network datasets. Specially, we provide a nation-wide social network dataset with individual characteristics. [Data]
  • Keywords: complex explanatory models, social network formation, game theory, node embedding, agent based modeling, diversity versus homophily
  • Coverage: PNAS Journal Club

[3] Yuan Yuan, Christos Nicolaides, Dean Eckles, and Alex Pentland, "Who Motivates More Workouts: Friends or Strangers? " working paper.
  • Abstract: Fitness enthusiasts suggest people walk 10,000 steps (approximately five miles) per day to stay healthy. Although many people are determined to exercise frequently, they may struggle to do so. Our study seeks to identify the effectiveness of a third incentive — prosocial incentive — on fitness behavior. We collaborate with WeChat, the largest Chinese social networking platform, to examine the effectiveness of prosocial incentives on healthy fitness behavior, and compare it with peer effects. Specifically, we examine WeRun, a sports mini-program on WeChat and its “step donation” feature. We use high-dimensional matching and a field experiment of 30 million users to identify the efficacy of the “step donation” feature.
  • Keywords: causal data mining, experimental design, prosocial behavior, health
  • [Slides][Video] for the observational part; presented at NetSci'20
  • Draft available upon request

[4] Yuan Yuan, Kristen Altenburger, and Farshad Kooti, "Causal Network Motifs: Identifying Heterogeneous Spillover Effects in A/B Tests", to appear in the Web Conference (WWW'2021) [preprint].
  • Abstract: Randomized experiments, or "A/B" tests, remain the gold standard for evaluating the causal effect of a policy intervention or product change. However, experiments in settings such as social networks, where users are interacting and influencing one another, violate classical assumptions of no interference for valid causal inference. Existing solutions include accounting for the fraction or count of treated neighbors in a user's network. Yet, there are often a high number of researcher degrees of freedom in specifying network interference conditions and most current methods do not account for the local network structure beyond just simple counts of the number of neighbors. To fill this gap, we provide an approach of accounting for both the local structure in a user's network via motifs as well as the assignment conditions of neighbors. We first introduce and employ "causal network motifs", i.e. network motifs that take into account the assignment conditions in local networks; and then we propose a tree-based algorithm for categorizing different network interference conditions and estimating their average potential outcomes. We test our method on a real-world experiment on a large-scale network and a synthetic network setting.
  • Keywords: causal data mining, experimental design, network interference

Talks

Invited talks
  • [5] “Causal Network Motifs: Identifying Heterogenous Spillover Effects in A/B Tests”, invited talk, Networks Seminar, Oxford Mathematical Institute, Nov 2020.
  • [4] “Causal identifications in observational studies” invited talk, Big Data and Social Computing, Aug 2020.
  • [3] “Identifying gift contagion in online groups, ” guest lecturer for Behavioral Economics, Tsinghua University (remotely), Mar 2020.
  • [2] “Predicting economic growth by social diversity, ” invited talk, International Conference on Social Computing, Aug 2019.
  • [1] “Trading off between homophily and social exchange for social network formation, ” invited talk, Beijing Normal University, Jan 2019.
Conference presentations
  • [15] “Network Motifs with Treatment Assignment Conditions: Identifying Heterogeneous Network Interference Eects in A/B Tests,” Conference on Digital Experimentation (CODE), Nov 2020.
  • [14] “Prosocial Incentives and Workouts: Evidence from a Massive Online Experiment,” Conference on Digital Experimentation (CODE), Nov 2020.
  • [13] “Who motivates more workouts: Friends or strangers?” International Conference on Network Science (NetSci), Rome, Sept 2020.
  • [12] “Who motivates more workouts: Friends or strangers?”, International Conference on Computational Social Science (IC2S2), Boston, July 2020.
  • [11] “The contagion of online gift giving,” INFORMS Annual Meeting, Seattle, Oct 2019.
  • [10] “Does prosocial contagion increase inequality? A large-scale online field experiment,” International
  • Conference on Computational Social Science (IC2S2), Amsterdam, July 2019.
  • [9] “A large-scale natural experiment of indirect reciprocity,” Conference on Digital Experimentation (CODE), Boston, Oct 2018.
  • [8] “A large-scale natural experiment of indirect reciprocity,” Advances in Field Experiments (AFE), Boston, Oct 2018.
  • [7] “Online red packets: A large-scale empirical study of gift giving on WeChat,” International Conference on Computational Social Science (IC2S2), Evanston, July 2018.
  • [6] “An interpretable approach for social network formation among heterogeneous agents,” International Conference on Computational Social Science (IC2S2), Evanston, July 2018.
  • [5] “Trade-off between social exchange and homophily in social network formation,” International Conference on Network Science (NetSci), Paris, June 2018.
  • [4] “A large-scale empirical study of gift giving on WeChat,” Annual Conference on Network Science and Economics (NetSci Econ), Nashville, Apr 2018.
  • [3] “Trade-off between social exchange and homophily in social network formation,” Annual Conference on Network Science and Economics (NetSci Econ), Nashville, Apr 2018.
  • [2] “Social network formation based on endowment exchange. and Social Representation,” Conference on Complex Systems (CCS), Cancún, Oct 2017.
  • [1] “Interpretable and effective opinion spam detection via temporal pattern mining across websites,” IEEE International Conference on Big Data (BigData), Washington DC, Dec 2016.

Services

Technical program committee: Conference ON Digital Experimentation (CODE), 2019 & 2020.

Organizer: Summer Institute in Computational Social Science (SICSS), Beijing .

Reviewer: Management Science, MIS Quarterly, IC2S2, CODE, TKDD, IEEE Transactions on Big Data, HKS Misinformation Review.


Notes

I occasionally share my thoughts or post blogs. I would appreciate thoughts or comments.


Contact

If you have any questions or just want to have a chat, please contact me via:

Email: yuanyuan [at] purdue [dot] edu

Purdue office: 403 W State St, Room 709, West Lafayette, IN 47907

Twitter: