Research

Working papers

1. Yuan Yuan and Kristen Altenburger, " A two-part machine learning approach to characterizing network interference in A/B testing" [tldr]
   - Conference version: Causal Network Motifs: Identifying Heterogeneous Spillover Effects in A/B Tests
   - Major revision at Manufacturing & Service Operations Management

   TLDR: This paper presents a machine learning approach to address network interference in A/B testing, where the outcome for one participant can affect others. It introduces "causal network motifs" to represent network structures and uses transparent machine learning models (like decision trees and nearest neighbors) to improve the accuracy of testing results. The approach is validated on large-scale experiments including one with 1-2 million Instagram users.

2. Yuan Yuan, Mengting Wan, Brent Hecht, Jaime Teevan, Longqi Yang " Large-scale telemetry data reveal associations between intra-organizational networks and firm financial performance, " invited to resubmit to Management Science. [tldr]

   TLDR: This study explores how the relationships within an organization (intra-organizational networks) correlates with its financial success by analyzing over 1500 companies' financial data and internal interactions over five quarters.

3. Shan Huang*, Chen Wang*, Yuan Yuan*, Jinglong Zhao*, Jingjing Zhang, " Estimating effects of long term treatments " [tldr]
   - Conference version accepted at ACM Economics and Computation (EC'2023)
   - Major revision at Management Science

   TLDR: This paper introduces a "longitudinal surrogate framework" to estimate long-term effects of treatments (like product updates or algorithm changes) using short-term A/B testing data. By leveraging user attributes, intermediate metrics, and treatments over time, the framework can predict long-term outcomes, reducing estimation bias significantly, as evidenced by real-world and synthetic experiments on the WeChat platform.

4. Yuan Yuan, Christos Nicolaides, and Dean Eckles, " Promoting physical activity through prosocial incentives on mobile platforms " [tldr]

   TLDR: The study explores the impact of prosocial incentives, specifically a “step donation” feature on a major social media platform, on promoting physical activity among users. In a large-scale field experiment, users who walked more than 10,000 steps could trigger charitable donations. Findings showed that nudges increased the likelihood of walking more than 10,000 steps by 16.4%, with an average increase of 2,056 steps. Achievement-based nudges were more effective than charity-based ones.

5. Yan Leng* and Yuan Yuan* " Do LLM Agents Exhibit Social Behavior?"[tldr]
   - Major revision at Information Systems Research
   
   
   

   TLDR: This paper investigates whether Large Language Models (LLMs) like GPT-4 can exhibit social behaviors similar to humans, focusing on aspects like social learning, preference, and cooperation. Using classical laboratory experiments adapted for LLMs, the study finds that GPT-4 shows human-like social behaviors including distributional preferences, reciprocity, and responsiveness to group identity, with notable differences in fairness preference and social learning strategies.

6. Marios Papachristou and Yuan Yuan " Network Formation and Dynamics Among Multi-LLMs " [tldr]

   TLDR: This study examines the behaviors and preferences of large language models (LLMs) in network formation, analyzing their impact on social and business interactions. It focuses on micro-level properties like preferential attachment, triadic closure, and homophily, as well as macro-level community structures and small-world phenomena. The research finds that LLMs can generate networks with these properties and that in real-world settings, LLMs prioritize homophily and triadic closure in their networking decisions, aiding in community structure reinforcement.

Journal Articles

1. Marie Charpignon*, Yuan Yuan*, Dehao Zhang*, Fereshteh Amini, Longqi Yang, Sonia Jaffe, Siddharth Suri "Navigating the new normal: Examining co-attendance in a hybrid work environment", Proceedings of the National Academy of Sciences, IF=11.1, 2023. [tldr]

   TLDR: A study of about 43,000 employees at a global tech company found that individuals were more likely to attend the office when their managers and teammates were present, with a 29% increase in attendance when managers were onsite. The research suggests that organic coordination of work schedules occurs, especially among new hires and those in corporate or operations roles, indicating that companies could enhance teamwork by promoting more coordinated in-office attendance through digital tools or scheduling strategies.

2. Yuan Yuan, Tracy Xiao Liu, Chenhao Tan, Qian Chen, Alex 'Sandy' Pentland, and Jie Tang, " Gift contagion in online groups: Evidence from Virtual red packets ", Management Science, 2023.[tldr]

   TLDR: This study examines gift contagion in online groups using 36 million data points from a social site in East Asia, showing that receiving gifts increases the likelihood of sending gifts, especially among those who receive the most. The research utilizes a natural experimental design to establish causality, finding that gift giving fosters group interaction and solidarity

3. Yuan Yuan, Eaman Jahani, Shengjia Zhao, Yong-Yeol Ahn, and Alex Pentland, " Implications of COVID-19 vaccination heterogeneity in mobility networks " , Communications Physics (Nature Portfolio), IF=6.5, 2023.
4. Ding Lyu, Yuan Yuan (corr author), Lin Wang, Xiaofan Wang, Alex Pentland, " Investigating and modeling the dynamics of long ties " , Communications Physics (Nature Portfolio), IF=6.5, 2022.
5. Yuan Yuan, Ahmad Alabdulkareem, and Alex 'Sandy' Pentland, " An interpretable approach for social network formation among heterogeneous agents", Nature Communications, IF=16.6, 2018. [tldr]

   TLDR: The paper introduces a model that blends game theory, machine learning, and sociology to understand social network formation, using "endowment vectors" to represent agents' features. It focuses on how agents form links based on a balance between exchange benefits and coordination costs, offering insights into the network's structure and dynamics.

Conference proceedings

1. David Gamba, Yulin Yu, Yuan Yuan, Grant Schoenebeck, Daniel M Romero, "Exit ripple effects: Understanding the disruption of socialization networks following employee departures", the Web Conference (WWW), acceptance rate=20.2%, 2024.
2. Yongkang Guo, Yuan Yuan, Jinshan Zhang, Yuqing Kong, Zhihua Zhou, Zheng Cai, "Near-Optimal Experimental Design Under the Budget Constraint in Online Platforms", the Web Conference (WWW), acceptance rate=19.2%, 2023.
3. Yuan Yuan, Kristen Altenburger, and Farshad Kooti, " Causal Network Motifs: Identifying Heterogeneous Spillover Effects in A/B Tests ", the Web Conference (WWW), acceptance rate=20.6%, 2021.
4. Yuan Yuan, Sihong Xie, Chun-Ta Lu, Jie Tang, and Philip S. Yu, " Interpretable and effective opinion spam detection via temporal pattern mining across websites ." IEEE International Conference on Big Data, acceptance rate=18.7%, 2016.

Other Publications

1. Yuan Yuan, Tracy Xiao Liu " Spillover effects in online field experiments: an annotated reading list ." ACM SIGecom Exchanges, 2022.

Work in Progress

1. Yuan Yuan " Multi-Calibrated heterogeneous treatment effect estimation "

* indicates equal contributions.
** Other projects follow the first-author-emphasis norm, which is common in fields like CS and natural sciences.

Services

Technical Program Committee: MIT Conference on Digital Experimentation, 2019-2021.

Reviewer: NSF Proposal, Journal of the American Statistical Association, Management Science, MIS Quarterly, Information Systems Research, ACM Transactions on Knowledge Discovery from Data, IEEE Transactions on Big Data, HBS Misinformation Review, American Causal Inference Conference (ACIC), IC2S2, WITS, CIST, WINE

Organizer: SICSS Beijing 2021, WINE Experimentation Workshop 2021.

Teaching

Social network analytics (undergrad, new course), Spring 2023, 2024

Introduction to Management Information Systems (undergrad), Spring 2022, 2024

Social network analytics (PhD, new course), Fall 2021

I will be teaching undergraduate level in Spring 2022.