This work examines information diffusion in decentralized social media platforms, specifically on Mastodon due to its federated architecture. Unlike centralized platforms like Twitter, Mastodon distributes content across independently operated yet interconnected servers, known as "instances." In this decentralized environment, inter-instance diffusion plays a critical role in shaping communication patterns. Using a dataset of Mastodon posts, or "toots," related to the Gaza conflict, we analyzed how user-, toot-, and instance-level characteristics drive cross-instance diffusion. Our findings demonstrate that these features collectively shape diffusion, with instance-level characteristics playing a particularly crucial role. This work contributes to the theories of decentralized communication by emphasizing the importance of federated architectures and inter-instance interactions. Practical implications include strategies for platform governance, crisis communication, and misinformation control, with a focus on the potential of decentralized platforms to balance localized interactions with global reach.
Health Comes at a Hidden Cost: Strategic Pricing and Equity in Heart Supplement Markets
Li Zeng and Ziqi Hang
In Proceedings of the 59th Hawaii International Conference on System Sciences (HICSS), 2026
Strategic pricing can improve profitability but may exacerbate disparities among vulnerable consumers. To examine this tension, this work utilizes NielsenIQ data on heart-health supplement purchases, analyzing empirical coupon usage and price disparities associated with household demographics and health status. Based on these models, a simulation study evaluates the impact of four pricing strategies: uniform, second-degree, third-degree, and combined, under varying behavioral assumptions. Findings highlight how revenue gains from demographic-based pricing can come at the cost of increased inequity, but also show that more balanced outcomes which support both profitability and fairness, are achievable through combined or behavior-based strategies.
The Cost of Inaccessibility: Retail Discrimination and Mobility-Constrained Consumers
Li Zeng and Ziqi Hang
In Proceedings of the 59th Hawaii International Conference on System Sciences (HICSS), 2026
This study investigates whether individuals with mobility-related health conditions face systematic disadvantages in retail pricing, both in physical stores and on online platforms. While price discrimination is a common feature of modern markets, limited attention has been paid to how physical mobility constraints may affect consumers’ ability to access lower prices or respond to promotions. Using matched observational data from the NielsenIQ and the Open E-Commerce dataset, we compare purchasing behavior and price outcomes between consumers with and without mobility-related health conditions. We find that mobility-constrained individuals pay modestly higher prices in physical stores and face even larger price disparities in online purchases, particularly among wheelchair users. Our findings highlight mobility as an underexamined axis of vulnerability in consumer markets. The results have implications for the design of retail pricing systems and digital platforms, as well as for policy efforts aimed at improving equity in access to essential goods.
2025
Real Name, Real Face, Real Talk? Anonymity and Toxicity on Mastodon
Krzysztof Wójcik, Li Zeng, and Sijia Ma
In Workshop Proceedings of the 19th International AAAI Conference on Web and Social Media, 2025
This study investigates the relationship between user anonymity and toxicity in discourse on Mastodon - a decentralized social media platform. We develop a profile-based anonymity classification framework using name and face presence as identity cues, and apply a pre-trained language model to detect toxicity in user-generated posts. Our results show that identifiable users, particularly those who disclose both name and face, tend to post less toxic content. We also find that higher levels of instance moderation are associated with reduced toxicity overall. These findings highlight how identity presentation and platform governance jointly shape discourse quality in decentralized social networks.
Sentiment Dynamics and Shifts across Instances on Mastodon
Seeun Kim, Li Zeng, Sijia Ma, and Giulia Sturlese
In Workshop Proceedings of the 19th International AAAI Conference on Web and Social Media, 2025
This study investigates how sentiment shifts unfold on Mastodon, a decentralized social network, during the 2023 Gaza conflict. We examine how the federated structure shapes the spread and transformation of sentiment in crisis-related discourse. We find that posts with broader reach tend to show more volatile sentiment, major conflict events coincide with sharp sentiment changes, and reply interactions often diverge in sentiment from the original post. These findings highlight how decentralized infrastructures mediate sentiment dynamics in online communications.
2022
How KOLs Influence Consumer Purchase Intention in Short Video Platforms: Mediating Effects of Emotional Response and Virtual Touch
Xuandi Gong, Jinluan Ren, Li Zeng, and Rubin Xing
International Journal of Information Systems in the Service Sector, 2022
In recent years, key opinion leader (KOL) marketing opens up a new mode of social commerce by effectively integrating social networking and marketing since it has been successfully taking advantage of KOL’s high popularity to promote products. This study expands the stimulus-organism-response (S-O-R) model by combining the communication persuasion theory with the flow experience theory. The model considers the characteristics of KOLs and published content features as independent variables, consumer perception as the mediating variable, and consumer purchase intention as the dependent variable. This study also refines the measurement dimensions of each variable and analyzes KOL impacts on consumer purchase intention on short video platforms. After analyzing 357 valid questionnaires, the authors find that the variables—reputation, perceived fit, aesthetic quality, and content richness—have significant impacts on consumer purchase intention where virtual touch and emotional response play intermediary roles. This study provides insights into KOL marketing.
2019
Detecting journalism in the age of social media: three experiments in classifying journalists on twitter
Li Zeng, Dharma Dailey, Owla Mohamed, Kate Starbird, and Emma S Spiro
In Proceedings of the International AAAI Conference on Web and Social Media, 2019
The widespread adoption of networked information and communications technologies (i.e. ICTs) blurs traditional boundaries between journalist and citizen. The role of the journalist is adapting to structural changes in the news industry and dynamic audience expectations. For researchers who seek to understand what, if any, distinct role journalists play in the production and propagation of breaking news, it is vital to be able to identify journalists in social media spaces. In many cases, this can be challenging due to the limited information and metadata about social media users. In this work, we use a supervised machine learning model to automatically distinguish journalists from non-journalists in social media spaces. Leveraging Twitter data collected from three crisis events of different types, we examine how profile information, social network structure, posting behavior and language distinguish journalists from others. Additionally, we evaluate how the performance of the journalist classification model varies by context (i.e. types of crisis events) and by journalism outlets (i.e. print versus broadcast journalism), and discuss challenges in automatic journalist detection. Implications of this work are discussed; in particular we argue for the value of such methods for scaling analysis in journalism studies beyond the capacity of human coders. Employing classification methods in this context allows for systematic, large-scale studies of the role of journalists online.
“Friending” in Online Fitness Communities: Exploring Activity-Based Online Network Structure
Li Zeng, Zack W. Almquist, and Emma S. Spiro
In Proceedings of the 52nd Hawaii International Conference on System Sciences, 2019
Individuals are influenced by both direct and indirect interaction with their social contacts. While peer influence is known to affect health-related outcomes such as exercise, limited work has fully explored how social networks are structured to support (or inhibit) interaction that could lead to positive health behaviors. With the development of pervasive technology and rise of personal health and wellness tracking, increasing attention has been paid to promoting positive fitness behaviors through social interaction mechanisms in online fitness communities. This trend offers a unique opportunity to understand the opportunity structures for personal health and wellness support. Utilizing a large-scale behavioral trace dataset from the online fitness community Strava, we examine how the size of people’s personal network is structured by demographics (e.g. gender and age) and an economic indicator (i.e. if they pay for a premium account). We employ stochastic process models to characterize the empirical network degree distributions in this population of fitness community members. We find that gender, age and account status are associated with distinct network structure. Results have implications in the analysis and the design of health interventions that make use of network relationships in online settings.
Unbiased sampling of users from (online) activity data
Zack W Almquist, Sakshi Arya, Li Zeng, and Emma Spiro
Online platforms offer new opportunities to study human behavior. However, while social scientists are often interested in using behavioral trace data—data created by a user over the course of their everyday life—to draw inferences about users, many online platforms only allow data to be sampled based on user activities (leading to data sets that are biased toward highly active users). Here, we introduce a simple method for reweighting activity-based sample statistics in order to provide descriptive (and potentially model-based) estimates of the user population. We illustrate these techniques by applying them to a case study of an online fitness community (Strava) and use it to explore basic network properties. Last, we explore the weights effect on model-based estimates for count data.
2018
Stay Connected and Keep Motivated: Modeling Activity Level of Exercise in an Online Fitness Community
Li Zeng, Zack W. Almquist, and Emma S. Spiro
In International Conference on Social Computing and Social Media, 2018
Recent years have witnessed a growing popularity of activity tracking applications. Previously work has focused on three major types of social interaction features in such applications: cooperation, competition and community. Such features motivate users to be more active in exercise and stay within the track of positive behavior change. Online fitness communities such as Strava encourage users to connect to peers and provide a rich set of social interaction features. Utilizing a large-scale behavioral trace data set, this work aims to analyze the dynamics of online fitness behaviors and network subscription as well as the relationship between them. Our results indicate that activeness of fitness behaviors not only has seasonal variations, but also vary by user group and how well users are connected in an online fitness community. These results provide important implications for studies on network-based health and design of application features for health promotion.
2017
Let’s Workout! Exploring Social Exercise in an Online Fitness Community
Increasing attention has been paid to promoting certain healthy habits through social interaction in online communities. At the intersection of social media and activity tracking applications, these platforms capture information on physical activities as well as peer-to-peer interactions. Importantly, they also offer researchers a novel opportunity to understand health behaviors by utilizing the large-scale behavioral trace data they archive. In this study we explore the characteristics and dynamics of social exercise (i.e. fitness activities with at least one peer physically co-present) using data collected from an online fitness community popular with cyclists and runners. In particular, we ask if factors such as temporal seasonality, activity performance and social feedback vary by the number of people participating in an activity; we do so by comparing associations for both men and women. Our results indicate that when peers are physically co-present for fitness activities (i.e. group workouts), exercise tends to be more intense and receive more feedback from other users, across both genders. Findings also suggest gender differences in the observed tendency to complete activities with others. These results have important implications for health and wellness interventions.
2016
#Unconfirmed: Classifying Rumor Stance in Crisis-Related Social Media Messages
Li Zeng, Kate Starbird, and Emma S. Spiro
In Proceedings of the 10th International AAAI Conference on Web and Social Media, 2016
It is well-established that within crisis-related communications, rumors are likely to emerge. False rumors, i.e. misinformation, can be detrimental to crisis communication and response; it is therefore important not only to be able to identify messages that propagate rumors, but also corrections or denials of rumor content. In this work, we explore the task of automatically classifying rumor stances expressed in crisisrelated content posted on social media. Utilizing a dataset of over 4,300 manually coded tweets, we build a supervised machine learning model for this task, achieving an accuracy over 88% across a diverse set of rumors of different types.
Rumors at the Speed of Light? Modeling the Rate of Rumor Transmission during Crisis
Li Zeng, Kate Starbird, and Emma S. Spiro
In Proceedings of the 49th Hawaii International Conference on System Sciences, 2016
Social media have become an established feature of the dynamic information space that emerges during crisis events. Both emergency responders and the public use these platforms to search for, disseminate, challenge, and make sense of information during crises. In these situations rumors also proliferate, but just how fast such information can spread is an open question. We address this gap, modeling the speed of information transmission to compare retransmission times across content and context features. We specifically contrast rumor-affirming messages with rumor-correcting messages on Twitter during a notable hostage crisis to reveal differences in transmission speed. Our work has important implications for the growing field of crisis informatics.
2013
A Design of Mobile Payments Business Model Based on Value Network
Jinluan Ren, Xiafei Zhuo, Li Zeng, and Bo Li
In Proceedings of the 20th International Conference on Industrial Engineering and Engineering Management, 2013
According to the characters of mobile payments and the requirements of business model design, some designing principles and essentials were proposed. Based on the process of value creation in mobile payments, the value network model of mobile payments was constructed. The design of business model of mobile payments was divided into four phases: value discovery, value creation, value management, and value realization. Every phase was designed in detail. At the end, the future research directions were proposed.