Amin Banayeeanzade, Ala N. Tak, Fatemeh Bahrani, Anahita Bolourani, Leonardo Blas, Emilio Ferrara, Jonathan Gratch, Sai Praneeth Karimireddy. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

This paper examines how player flexibility – a player’s willingness to engage in a breadth of options or specialize – manifests across two gaming environments: League of Legends (League) and Teamfight Tactics (TFT). We analyze the gameplay decisions of 4,830 players who have played at least 50 competitive games in both titles and explore cross-game dynamics of behavior retention and consistency. Our work introduces a novel cross-game analysis that tracks the same players’ behavior across two different environments, reducing self-selection bias. Our findings reveal that while games incentivize different behaviors (specialization in League versus flexibility in TFT) for performance-based success, players exhibit consistent behavior across platforms. This study contributes to long-standing debate about agency versus structure, showing individual agency may be more predictive of cross-platform behavior than game-imposed structure in competitive settings. These insights offer implications for game developers, designers and researchers interested in building systems to promote behavior change.

STRIVE is a taxonomy for multiplayer games organized around player behavior features such as sociality, communication, and experience. We extend STRIVE from Sky: Children of the Light, a relationship-focused social game, to World of Tanks (WoT), a game centered on team combat, player clans, and competitive performance. Using three player data snapshots from 2020 and the original STRIVE framework, we recover four stable player types in WoT: Veterans, Socialites, Squad Players, and Newbies. These segments persist across time and align with the original STRIVE dimensions, while adding a WoT-specific performance dimension. Prediction experiments illustrate that future battle participation varies in predictability across player types: Veterans and Squad Players are consistently more predictable than Socialites and Newbies. Together, these findings support STRIVE as a generalizable framework for comparing social play across multiple games.

Generative agents are rapidly advancing in sophistication, raising urgent questions about how they might coordinate when deployed in online ecosystems. This is particularly consequential in information operations (IOs), influence campaigns that aim to manipulate public opinion on social media. While traditional IOs have been orchestrated by human operators and relied on manually crafted tactics, agentic AI promises to make campaigns more automated, adaptive, and difficult to detect. This work presents the first systematic study of emergent coordination among generative agents in simulated IO campaigns. Using generative agent-based modeling, we instantiate IO and organic agents in a simulated environment and evaluate coordination across operational regimes, from simple goal alignment to team knowledge and collective decision-making. As operational regimes become more structured, IO networks become denser and more clustered, interactions more reciprocal and positive, narratives more homogeneous, amplification more synchronized, and hashtag adoption faster and more sustained. Remarkably, simply revealing to agents which other agents share their goals can produce coordination levels nearly equivalent to those achieved through explicit deliberation and collective voting. Overall, we show that generative agents, even without human guidance, can reproduce coordination strategies characteristic of real-world IOs, underscoring the societal risks posed by increasingly automated, self-organizing IOs.

Social media platforms have become vital spaces for public discourse, serving as modern agorás where a wide range of voices influence societal narratives. However, their open nature also makes them vulnerable to exploitation by malicious actors, including state-sponsored entities, who can conduct information operations (IOs) to manipulate public opinion. The spread of misinformation, false news, and misleading claims threatens democratic processes and societal cohesion, making it crucial to develop methods for the timely detection of inauthentic activity to protect the integrity of online discourse. In this work, we introduce a methodology designed to identify users orchestrating information operations, a.k.a. IO drivers, across various influence campaigns. Our framework, named IOHunter, leverages the combined strengths of Language Models and Graph Neural Networks to improve generalization in supervised, scarcely-supervised, and cross-IO contexts. Our approach achieves state-of-the-art performance across multiple sets of IOs originating from six countries, significantly surpassing existing approaches. This research marks a step toward developing Graph Foundation Models specifically tailored for the task of IO detection on social media platforms.

We introduce STRIVE, a framework to build socio-behavioral taxonomies in multiplayer online games using unsupervised learning on common features across many social games.This work demonstrates this framework on "Sky: Children of the Light," a social adventure game by Thatgamecompany, using features such as cooperative play, social bonding, and in-game communication.After performing descriptive statistics, clustering and dimensionality reduction, we assign semantic categories to these behavior clusters.Next we perform a behavior prediction experiment where the most social cluster's behavior has a higher correlation with future play time and chats sent than predicting on the full dataset.These results suggest the importance of customized, or personalized, player behavior prediction models.Moreover, this framework could be easily extended to other games and further augment our understanding of human behavior in virtual worlds, ultimately aiding social scientists and game designers to better match players together for healthier online interactions.

Understanding susceptibility to online influence is crucial for mitigating the spread of misinformation and protecting vulnerable audiences. This paper investigates susceptibility to influence within social networks, focusing on the differential effects of influence-driven versus spontaneous behaviors on user content adoption. Our analysis reveals that influence-driven adoption exhibits high homophily, indicating that individuals prone to influence often connect with similarly susceptible peers, thereby reinforcing peer influence dynamics, whereas spontaneous adoption shows significant but lower homophily. Additionally, we extend the Generalized Friendship Paradox to influence-driven behaviors, demonstrating that users' friends are generally more susceptible to influence than the users themselves, de facto establishing the notion of Susceptibility Paradox in online social influence. This pattern does not hold for spontaneous behaviors, where friends exhibit fewer spontaneous adoptions. We find that susceptibility to influence can be predicted using friends' susceptibility alone, while predicting spontaneous adoption requires additional features, such as user metadata. These findings highlight the complex interplay between user engagement and characteristics in spontaneous content adoption. Our results provide new insights into social influence mechanisms and offer implications for designing more effective moderation strategies to protect vulnerable audiences.

Coordinated information operations remain a persistent challenge on social media, despite platform efforts to curb them. While previous research has primarily focused on identifying these operations within individual platforms, this study shows that coordination frequently transcends platform boundaries. Leveraging newly collected data of online conversations related to the 2024 U.S. Election across 𝕏 (formerly, Twitter), Facebook, and Telegram, we construct similarity networks to detect coordinated communities exhibiting suspicious sharing behaviors within and across platforms. Proposing an advanced coordination detection model, we reveal evidence of potential foreign interference, with Russian-affiliated media being systematically promoted across Telegram and 𝕏. Our analysis also uncovers substantial intra- and cross-platform coordinated inauthentic activity, driving the spread of highly partisan, low-credibility, and conspiratorial content. These findings highlight the urgent need for regulatory measures that extend beyond individual platforms to effectively address the growing challenge of cross-platform coordinated influence campaigns.

The present research on team collaboration is typically performed through qualitative interview based studies or social network measurements of connectedness through co-play. In this study, we take the unique approach to build networks from direct messages between players in the massive online game World of Tanks where players self-organize into clans with specific roles assigned from military rankings (from Private to Commander). We explore the relationship between team communication volume and skill level, the impact of communication features on clan rating, and the differences in communication hierarchy between high and low-rated clans. Our findings reveal that higher-rated clans send more pre-battle chat messages, suggesting that effective communication and strategic planning are key to team performance. Evidence shows teams who use voice chat during battle are significantly higher ranked. Finally, we reveal that the highest rated clans have more connected lower-ranked members emphasizing that these teams are ''only as strong as their weakest link.'' This research is guided by the Transactive Memory Systems and Collective Intelligence theories which serve to expand the contribution of this research outside of games to other forms of virtual collaboration.

Alexander Spangher, Serdar Tumgoren, Ben Welsh, Nanyun Peng, Emilio Ferrara, Jonathan May. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

This study investigates the concept of flexibility within League of Legends, a popular online multiplayer game, focusing on the relationship between user adaptability and team success. Utilizing a dataset encompassing players of varying skill levels and play styles, we calculate two measures of flexibility for each player: overall flexibility and temporal flexibility. Our findings suggest that the flexibility of a user is dependent upon a user’s preferred play style, and flexibility does impact match outcome. This work also shows that skill level not only indicates how willing a player is to adapt their play style but also how their adaptability changes over time. This paper highlights the duality and balance of specialization versus flexibility, providing insights that can inform strategic planning, collaboration and resource allocation in competitive environments.

Hyundong Cho, Shuai Liu, Taiwei Shi, Darpan Jain, Basem Rizk, Yuyang Huang, Zixun Lu, Nuan Wen, Jonathan Gratch, Emilio Ferrara, Jonathan May. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

Misinformation proliferation on social media platforms is a pervasive threat to the integrity of online public discourse. Genuine users, susceptible to others' influence, often unknowingly engage with, endorse, and re-share questionable pieces of information, collectively amplifying the spread of misinformation. In this study, we introduce an empirical framework to investigate users' susceptibility to influence when exposed to unreliable and reliable information sources. Leveraging two datasets on political and public health discussions on Twitter, we analyze the impact of exposure on the adoption of information sources, examining how the reliability of the source modulates this relationship. Our findings provide evidence that increased exposure augments the likelihood of adoption. Users tend to adopt low-credibility sources with fewer exposures than high-credibility sources, a trend that persists even among non-partisan users. Furthermore, the number of exposures needed for adoption varies based on the source credibility, with extreme ends of the spectrum (very high or low credibility) requiring fewer exposures for adoption. Additionally, we reveal that the adoption of information sources often mirrors users' prior exposure to sources with comparable credibility levels. Our research offers critical insights for mitigating the endorsement of misinformation by vulnerable users, offering a framework to study the dynamics of content exposure and adoption on social media platforms.

Training ML models which are fair across different demographic groups is of critical importance due to the increased integration of ML in crucial decision-making scenarios such as healthcare and recruitment. Federated learning has been viewed as a promising solution for collaboratively training machine learning models among multiple parties while maintaining their local data privacy. However, federated learning also poses new challenges in mitigating the potential bias against certain populations (e.g., demographic groups), as this typically requires centralized access to the sensitive information (e.g., race, gender) of each datapoint. Motivated by the importance and challenges of group fairness in federated learning, in this work, we propose FairFed, a novel algorithm for fairness-aware aggregation to enhance group fairness in federated learning. Our proposed approach is server-side and agnostic to the applied local debiasing thus allowing for flexible use of different local debiasing methods across clients. We evaluate FairFed empirically versus common baselines for fair ML and federated learning and demonstrate that it provides fairer models, particularly under highly heterogeneous data distributions across clients. We also demonstrate the benefits of FairFed in scenarios involving naturally distributed real-life data collected from different geographical locations or departments within an organization.

News articles are driven by the informational sources journalists use in reporting.Modeling when, how and why sources get used together in stories can help us better understand the information we consume and even help journalists with the task of producing it.In this work, we take steps toward this goal by constructing the largest and widest-ranging annotated dataset, to date, of informational sources used in news writing.We first show that our dataset can be used to train high-performing models for information detection and source attribution.Then, we introduce a novel task, source prediction, to study the compositionality of sources in news articles -i.e.how they are chosen to complement each other.We show good modeling performance on this task, indicating that there is a pattern to the way different sources are used together in news storytelling.This insight opens the door for a focus on sources in narrative science (i.e.planningbased language generation) and computational journalism (i.e. a source-recommendation system to aid journalists writing stories).1

On February 24, 2022, Russia invaded Ukraine. In the days that followed, reports kept flooding in from laymen to news anchors of a conflict quickly escalating into war. Russia faced immediate backlash and condemnation from the world at large. While the war continues to contribute to an ongoing humanitarian and refugee crisis in Ukraine, a second battlefield has emerged in the online space, both in the use of social media to garner support for both sides of the conflict and also in the context of information warfare. In this paper, we present a collection of nearly half a billion tweets, from February 22, 2022, through January 8, 2023, that we are publishing for the wider research community to use. This dataset can be found at https://github.com/echen102/ukraine-russia. Our preliminary analysis on a subset of our dataset already shows evidence of public engagement with Russian state-sponsored media and other domains that are known to push unreliable information towards the beginning of the war; the former saw a spike in activity on the day of the Russian invasion, while the other saw spikes in engagement within the first month of the war. Our hope is that this public dataset can help the research community to further understand the ever-evolving role that social media plays in information dissemination, influence campaigns, grassroots mobilization, and much more, during a time of conflict.

Estimating the political leanings of social media users is a challenging and ever more pressing problem given the increase in social media consumption. We introduce Retweet-BERT, a simple and scalable model to estimate the political leanings of Twitter users. Retweet-BERT leverages the retweet network structure and the language used in users' profile descriptions. Our assumptions stem from patterns of networks and linguistics homophily among people who share similar ideologies. Retweet-BERT demonstrates competitive performance against other state-of-the-art baselines, achieving 96%-97% macro-F1 on two recent Twitter datasets (a COVID-19 dataset and a 2020 United States presidential elections dataset). We also perform manual validation to validate the performance of Retweet-BERT on users not in the training data. Finally, in a case study of COVID-19, we illustrate the presence of political echo chambers on Twitter and show that it exists primarily among right-leaning users. Our code is open-sourced and our data is publicly available.

Social media provide a fertile ground where conspiracy theories and radical ideas can flourish, reach broad audiences, and sometimes lead to hate or violence beyond the online world itself. QAnon represents a notable example of a political conspiracy that started out on social media but turned mainstream, in part due to public endorsement by influential political figures. Nowadays, QAnon conspiracies often appear in the news, are part of political rhetoric, and are espoused by significant swaths of people in the United States. It is therefore crucial to understand how such a conspiracy took root online, and what led so many social media users to adopt its ideas. In this work, we propose a framework that exploits both social interaction and content signals to uncover evidence of user radicalization or support for QAnon. Leveraging a large dataset of 240M tweets collected in the run-up to the 2020 US Presidential election, we define and validate a multivariate metric of radicalization. We use that to separate users in distinct, naturally-emerging, classes of behaviors associated with radicalization processes, from self-declared QAnon supporters to hyper-active conspiracy promoters. We also analyze the impact of Twitter's moderation policies on the interactions among different classes: we discover aspects of moderation that succeed, yielding a substantial reduction in the endorsement received by hyperactive QAnon accounts. But we also uncover where moderation fails, showing how QAnon content amplifiers are not deterred or affected by the Twitter intervention. Our findings refine our understanding of online radicalization processes, reveal effective and ineffective aspects of moderation, and call for the need to further investigate the role social media play in the spread of conspiracies.

Win prediction is crucial to understanding skill modeling, teamwork and matchmaking in esports. In this paper we propose GCN-WP, a semi-supervised win prediction model for esports based on graph convolutional networks. This model learns the structure of an esports league over the course of a season (1 year) and makes predictions on another similar league. This model integrates over 30 features about the match and players and employs graph convolution to classify games based on their neighborhood. Our model achieves state-of-the-art prediction accuracy when compared to machine learning or skill rating models for LoL. The framework is generalizable so it can easily be extended to other multiplayer online games.

Malicious accounts spreading misinformation has led to widespread false and misleading narratives in recent times, especially during the COVID-19 pandemic, and social media platforms struggle to eliminate these contents rapidly. This is because adapting to new domains requires human intensive fact-checking that is slow and difficult to scale. To address this challenge, we propose to leverage news-source credibility labels as weak labels for social media posts and propose model-guided refinement of labels to construct large-scale, diverse misinformation labeled datasets in new domains. The weak labels can be inaccurate at the article or social media post level where the stance of the user does not align with the news source or article credibility. We propose a framework to use a detection model self-trained on the initial weak labels with uncertainty sampling based on entropy in predictions of the model to identify potentially inaccurate labels and correct for them using self-supervision or relabeling. The framework will incorporate social context of the post in terms of the community of its associated user for surfacing inaccurate labels towards building a large-scale dataset with minimum human effort. To provide labeled datasets with distinction of misleading narratives where information might be missing significant context or has inaccurate ancillary details, the proposed framework will use the few labeled samples as class prototypes to separate high confidence samples into false, unproven, mixture, mostly false, mostly true, true, and debunk information. The approach is demonstrated for providing a large-scale misinformation dataset on COVID-19 vaccines.

Understanding social interactions and generous behaviors have long been of considerable interest in the social sciences community. While the contagion of generosity is documented in the real world, less is known about such phenomenon in virtual worlds and whether it has an actionable impact on user behavior and retention. In this work, we analyze social dynamics in the virtual world of the popular massively multiplayer online role-playing game (MMORPG) Sky: Children of Light. We develop a framework to reveal the patterns of generosity in such social environments and provide empirical evidence of social contagion and contagious generosity. Players become more engaged in the game after playing with others and especially with friends. We also find that players who experience generosity first-hand or even observe other players conduct generous acts become more generous themselves in the future. Additionally, we show that both receiving and observing generosity lead to higher future engagement in the game. Since Sky resembles the real world from a social play aspect, the implications of our findings also go beyond this virtual world.

Affective computing is broadly applied to decision making systems ranging from mental health assessment to employability evaluation. The heterogeneity of human behavioral data poses challenges for both model validity and fairness. The limited access to sensitive attributes (e,g., race, gender) in real-world settings makes it more difficult to mitigate the unfairness of the model outcomes. In this work, we focus on the heterogeneity of human behavioral signals and analyze its impact on model fairness. We design a novel method named multi-layer factor analysis to automatically identify the heterogeneity patterns in high-dimensional behavioral data and propose a framework to enhance fairness of behavioral modeling without accessing sensitive attributes.

The popularity of online gaming has grown dramatically, driven in part by streaming and the billion dollar e-sports industry. Online games regularly update their software to fix bugs, add functionality that improve the game’s look and feel, and change the game mechanics to keep the games fun and challenging. An open question, however, is the impact of these changes on player performance and game balance, as well as how players adapt to these sudden changes. To address these questions, we use causal inference to measure the impact of software patches to League of Legends, a popular team-based multiplayer online game. We show that game patches have substantially different impacts on players, depending on their skill level and whether they take breaks between games. We find that the gap between good and bad players increases after a patch, despite efforts to make gameplay more equal. Moreover, longer between-game breaks tend to improve performance after patches. Overall, our results highlight the utility of causal inference, and specifically heterogeneous treatment effect estimation, as a tool to quantify the complex mechanisms of game balance and its interplay with players’ performance.

Disinformation campaigns on social media, involving coordinated activities from malicious accounts towards manipulating public opinion, have become increasingly prevalent. Existing approaches to detect coordinated accounts either make very strict assumptions about coordinated behaviours, or require part of the malicious accounts in the coordinated group to be revealed in order to detect the rest. To address these drawbacks, we propose a generative model, AMDN-HAGE (Attentive Mixture Density Network with Hidden Account Group Estimation) which jointly models account activities and hidden group behaviours based on Temporal Point Processes (TPP) and Gaussian Mixture Model (GMM), to capture inherent characteristics of coordination which is, accounts that coordinate must strongly influence each other's activities, and collectively appear anomalous from normal accounts. To address the challenges of optimizing the proposed model, we provide a bilevel optimization algorithm with theoretical guarantee on convergence. We verified the effectiveness of the proposed method and training algorithm on real-world social network data collected from Twitter related to coordinated campaigns from Russia's Internet Research Agency targeting the 2016 U.S. Presidential Elections, and to identify coordinated campaigns related to the COVID-19 pandemic. Leveraging the learned model, we find that the average influence between coordinated account pairs is the highest. On COVID-19, we found coordinated group spreading anti-vaccination, anti-masks conspiracies that suggest the pandemic is a hoax and political scam.

Modern protests are not limited to on-the-ground operations, and the ease and speed at which users can upload images to social media platforms has enabled protests to manifest online. Previous analysis of protest imagery from social media sites categorized these images into groups including texts, screenshots, memes, and artwork. However, large-scale manual annotation to identify different types of images is not feasible. By applying machine learning to a large Twitter dataset focused on the Stop Asian Hate movement, we found the type of image an account posted during protests on Twitter is tied to the credibility and political leaning of posted content, type of witnessing (remote or connective), and community formation.

Given: (a) clean training dataset, (b) trained graph machine learning pipeline, and (c) noisy test dataset; how do we make the pipeline robust to such noise? In an attempt to answer this question, we propose a modification to the pipeline that exploits both the content addressability of a restricted Boltzmann machine and the message passing capabilities of a graph neural network . We investigate which of the hidden layers of the pipeline is best suited for our proposed modification. By establishing that a decrease in mutual information indicates an increase in prediction accuracy which is in turn driven by progressive geometric clustering of the samples belonging to the same class, we attempt to explain the behaviour exhibited by the modified pipeline. Empirical results demonstrate that our approach makes the pipeline robust to some real-world noisy scenarios in citation datasets.

Although player performance in online games has been widely studied, few studies have considered the behavioral preferences of players and how they impact performance. In a competitive setting where players must cooperate with temporary teammates, it is even more crucial to understand how differences in playing style contribute to teamwork. Drawing on theories of individual behavior in teams, we describe a methodology to empirically profile players based on the diversity and conformity of their gameplay styles. Applying this approach to a League of Legends dataset, we find three distinct types of players that align with our theoretical framework: generalists, specialists, and mavericks. Importantly, the behavior of each player type remains stable despite players becoming more experienced. Additionally, we extensively investigate the benefits and drawbacks of each type of player by evaluating their individual performance, contribution to the team, and adaptation to changes in the game environment. We find that, overall, specialists tend to outperform others, while mavericks bear high risk but also potentially reap great rewards. Generalists are the most resilient to instability in the environment (game patches). We discuss the implications of these findings in terms of game design and community management, as well as team building in environments with varying levels of stability.

Social ties are the invisible glue that keeps together human ecosystems. Despite the massive amount of research studying the role of social ties in communities (groups, teams, etc.) and society at large, little attention has been devoted to study their interplay with other human behavioral dynamics. Of particular interest is the influence that social ties have on human performance in collaborative team-based settings. Our research aims to elucidate the influence of social ties on individual and team performance dynamics. We will focus on a popular multiplayer online battle arena (MOBA) collaborative team-based game,Defense of the Ancients 2 (Dota 2), a rich data set with millions of players and matches. Our research reveals that, when playing with their friends, individuals are systematically more active in the game as opposed to taking part in a team of strangers. However, we find that increased activity does not homogeneously lead to an improvement in players’ performance. Despite being beneficial to low-skill players, playing with friends negatively affects the performance of high-skill players. Our findings shed light on the mixed influence of social ties on performance and can inform new perspectives on virtual team management and behavioral incentives.

Machine learning models are at the foundation of modern society. Accounts of unfair models penalizing subgroups of a population have been reported in domains including law enforcement, job screening, etc. Unfairness can spur from biases in the training data, as well as from class imbalance, i.e., when a sensitive group's data is not sufficiently represented. Under such settings, balancing techniques are commonly used to achieve better prediction performance, but their effects on model fairness are largely unknown. In this paper, we first illustrate the extent to which common balancing techniques exacerbate unfairness in real-world data. Then, we propose a new method, called fair class balancing, that allows to enhance model fairness without using any information about sensitive attributes. We show that our method can achieve accurate prediction performance while concurrently improving fairness.

First identified in Wuhan, China, in December 2019, the outbreak of COVID-19 has been declared as a global emergency in January, and a pandemic in March 2020 by the World Health Organization (WHO). Along with this pandemic, we are also experiencing an "infodemic" of information with low credibility such as fake news and conspiracies. In this work, we present ReCOVery, a repository designed and constructed to facilitate research on combating such information regarding COVID-19. We first broadly search and investigate ~2,000 news publishers, from which 60 are identified with extreme [high or low] levels of credibility. By inheriting the credibility of the media on which they were published, a total of 2,029 news articles on coronavirus, published from January to May 2020, are collected in the repository, along with 140,820 tweets that reveal how these news articles have spread on the Twitter social network. The repository provides multimodal information of news articles on coronavirus, including textual, visual, temporal, and network information. The way that news credibility is obtained allows a trade-off between dataset scalability and label accuracy. Extensive experiments are conducted to present data statistics and distributions, as well as to provide baseline performances for predicting news credibility so that future methods can be compared. Our repository is available at http://coronavirus-fakenews.com.

Different social media environments enable different levels of community connectedness, which in turn affects the information that a user is exposed to. In this study, we create an agent-based model to investigate how the different levels of connectedness as well as network structure affects a group's diversity of opinions. In the model, agents are tasked with “liking” or “disliking” a set of objects. At each turn, each agent sees the most popular object amongst the agents that they are connected to. There are two main findings: 1) low network connectivity leads to more diversity amongst agents, 2) a complete network leads information popularity distribution to be more skewed, 3) the more random a network is, the more skewed information popularity distribution is. These findings suggest that online platforms that either create well-connected communities or present aggregated information of users (such as Billboard rankings) may lead to homogeneity in the subjective preference of the users over time.

Both offline and online human behaviors are affected by personality. Of special interests are online games, where players have to impersonate specific roles and their behaviors are extensively tracked by the game. In this paper, we propose to study the relationship between players' personality and game behavior in League of Legends (LoL), one of the most popular Multiplayer Online Battle Arena (MOBA) games. We use linear mixed effects (LME) models to describe relationships between players' personality traits (measured by the Five Factor Model) and two major aspects of the game: the impersonated roles and in-game actions. On the one hand, we study relationships within the game environment by modeling role attributes from match behaviors and vice versa. On the other hand, we analyze the relationship between a player's five personality traits and their game behavior by showing significant correlations between each personality trait and the set of corresponding behaviors. Our findings suggest that personality and behavior are highly entangled and provide a new perspective to understand how personality can affect behavior in role-based online games.

Forecasting of geopolitical events is a notoriously difficult task, with experts failing to significantly outperform a random baseline across many types of forecasting events. One successful way to increase the performance of forecasting tasks is to turn to crowdsourcing: leveraging many forecasts from non-expert users. Simultaneously, advances in machine learning have led to models that can produce reasonable, although not perfect, forecasts for many tasks. Recent efforts have shown that forecasts can be further improved by ``hybridizing'' human forecasters: pairing them with the machine models in an effort to combine the unique advantages of both. In this demonstration, we present Synergistic Anticipation of Geopolitical Events (SAGE), a platform for human/computer interaction that facilitates human reasoning with machine models.

How does the number of collaborators affect individual productivity? Results of prior research have been conflicting, with some studies reporting an increase in individual productivity as the number of collaborators grows, while other studies showing that the free-rider effect skews the effort invested by individuals, making larger groups less productive. The difference between these schools of thought is substantial: if a super-scaling effect exists, as suggested by former studies, then as groups grow, their productivity will increase even faster than their size, super-linearly improving their efficiency. We address this question by studying two planetary-scale collaborative systems: GitHub and Wikipedia. By analyzing the activity of over 2 million users on these platforms, we discover that the interplay between group size and productivity exhibits complex, previously-unobserved dynamics: the productivity of smaller groups scales super-linearly with group size, but saturates at larger sizes. This effect is not an artifact of the heterogeneity of productivity: the relation between group size and productivity holds at the individual level. People tend to do more when collaborating with more people. We propose a generative model of individual productivity that captures the non-linearity in collaboration effort. The proposed model is able to explain and predict group work dynamics in GitHub and Wikipedia by capturing their maximally informative behavioral features, and it paves the way for a principled, data-driven science of collaboration.

Until recently, social media was seen to promote democratic discourse on social and political issues. However, this powerful communication platform has come under scrutiny for allowing hostile actors to exploit online discussions in an attempt to manipulate public opinion. A case in point is the ongoing U.S. Congress investigation of Russian interference in the 2016 U.S. election campaign, with Russia accused of, among other things, using trolls (malicious accounts created for the purpose of manipulation) and bots (automated accounts) to spread misinformation and politically biased information. In this study, we explore the effects of this manipulation campaign, taking a closer look at users who re-shared the posts produced on Twitter by the Russian troll accounts publicly disclosed by U.S. Congress investigation. We collected a dataset with over 43 million elections-related posts shared on Twitter between September 16 and November 9, 2016 by about 5.7 million distinct users. This dataset includes accounts associated with the identified Russian trolls. We use label propagation to infer the users' ideology based on the news sources they shared, to classify a large number of them as liberal or conservative with precision and recall above 90%. Conservatives retweeted Russian trolls significantly more often than liberals and produced 36 times more tweets. Additionally, most of the troll content originated in, and was shared by users from Southern states. Using state-of-the-art bot detection techniques, we estimated that about 4.9% and 6.2% of liberal and conservative users respectively were bots. Text analysis on the content shared by trolls reveals that they had a mostly conservative, pro-Trump agenda. Although an ideologically broad swath of Twitter users were exposed to Russian trolls in the period leading up to the 2016 U.S. Presidential election, it was mainly conservatives who helped amplify their message.

According to the Center for Disease Control and Prevention, hundreds of thousands initiate smoking each year, and millions live with smoking-related diseases in the United States. Many tobacco users discuss their opinions, habits and preferences on social media. This work conceptualizes a framework for targeted health interventions to inform tobacco users about the consequences of tobacco use. We designed a Twitter bot named Notobot (short for No-Tobacco Bot) that leverages machine learning to identify users posting pro-tobacco tweets and select individualized interventions to curb their tobacco use. We searched the Twitter feed for tobacco-related keywords and phrases, and trained a convolutional neural network using over 4,000 tweets manually labeled as either pro-tobacco or not pro-tobacco. This model achieved a 90% accuracy rate on the training set and 74% on test data. Users posting protobacco tweets were matched with former smokers with similar interests who posted anti-tobacco tweets. Algorithmic matching, leveraging the power of peer influence, allows for the systematic delivery of personalized interventions based on real anti-tobacco tweets from former smokers. Experimental evaluation suggested that our system would perform well if deployed.

Network embedding, which aims to learn low-dimensional representations of nodes, has been used for various graph related tasks including visualization, link prediction, and node classification. Most existing embedding methods rely solely on network structure. However, in practice, we often have auxiliary information about the nodes and/or their interactions, e.g., the content of scientific papers in coauthorship networks, or topics of communication in Twitter mention networks. Here, we propose a novel embedding method that uses both network structure and edge attributes to learn better network representations. Our method jointly minimizes the reconstruction error for higher order node neighborhood, social roles, and edge attributes using a deep architecture that can adequately capture highly nonlinear interactions. We demonstrate the efficacy of our model over existing state-of-the-art methods on a variety of real-world networks including collaboration networks and social networks. We also observe that using edge attributes to inform network embedding yields better performance in downstream tasks such as link prediction and node classification.

Fashion is a multi-billion dollar industry with social and economic implications worldwide. To gain popularity, brands want to be represented by the top popular models. As new faces are selected using stringent (and often criticized) aesthetic criteria, a priori predictions are made difficult by information cascades and other fundamental trend-setting mechanisms. However, the increasing usage of social media within and without the industry may be affecting this traditional system. We therefore seek to understand the ingredients of success of fashion models in the age of Instagram. Combining data from a comprehensive online fashion database and the popular mobile image-sharing platform, we apply a ma- chine learning framework to predict the tenure of a cohort of new faces for the 2015 Spring/Summer season throughout the subsequent 2015-16 Fall / Winter season. Our framework successfully predicts most of the new popular models who appeared in 2015. In particular, we find that a strong social media presence may be more important than being under contract with a top agency, or than the aesthetic standards sought after by the industry.

With the emergence of social networking services, researchers enjoy the increasing availability of large-scale heterogenous datasets capturing online user interactions and behaviors. Traditional analysis of techno-social systems data has focused mainly on describing either the dynamics of social interactions, or the attributes and behaviors of the users. However, overwhelming empirical evidence suggests that the two dimensions affect one another, and therefore they should be jointly modeled and analyzed in a multi-modal framework. The benefits of such an approach include the ability to build better predictive models, leveraging social network information as well as user behavioral signals. To this purpose, here we propose the Constrained Latent Space Model (CLSM), a generalized framework that combines Mixed Membership Stochastic Blockmodels (MMSB) and Latent Dirichlet Allocation (LDA) incorporating a constraint that forces the latent space to concurrently describe the multiple data modalities. We derive an efficient inference algorithm based on Variational Expectation Maximization that has a computational cost linear in the size of the network, thus making it feasible to analyze massive social datasets. We validate the proposed framework on two problems: prediction of social interactions from user attributes and behaviors, and behavior prediction exploiting network information. We perform experiments with a variety of multi-modal social systems, spanning location-based social networks (Gowalla), social media services (Instagram, Orkut), e-commerce and review sites (Amazon, Ciao), and finally citation networks (Cora). The results indicate significant improvement in prediction accuracy over state of the art methods, and demonstrate the flexibility of the proposed approach for addressing a variety of different learning problems commonly occurring with multi-modal social data.

We introduce Cloud DIKW (Data, Information, Knowledge, Wisdom) as an analysis environment supporting scientific discovery through integrated parallel batch and streaming processing, and apply it to one representative domain application: social media data stream clustering. In this context, recent work demonstrated that high-quality clusters can be generated by representing the data points using high-dimensional vectors that reflect textual content and social network information. However, due to the high cost of similarity computation, sequential implementations of even single-pass algorithms cannot keep up with the speed of real-world streams. This paper presents our efforts in meeting the constraints of realtimesocial media stream clustering through parallelization in Cloud DIKW. Specifically, we focus on two system-level issues. Firstly, most stream processing engines such as Apache Storm organize distributed workers in the form of a directed acyclic graph (DAG), which makes it difficult to dynamically synchronize the state of parallel clustering workers. We tackle this challenge by creating a separate synchronization channel using a pub-sub messaging system (ActiveMQ in our case). Secondly, due to the sparsity of the high-dimensional vectors, the size of cancroids grows quickly as new data points are assigned tithe clusters. As a result, traditional synchronization that directly broadcasts cluster cancroids becomes too expensive and limits the scalability of the parallel algorithm. We address this problem by communicating only dynamic changes of the clusters rather than the whole centred vectors. Our algorithm under Cloud DIKWcan process the Twitter 10% data stream ("gardenhose") in realtimewith 96-way parallelism. By natural improvements to CloudDIKW, including advanced collective communication techniques developed in our Harp project, we will be able to process the full Twitter data stream in real-time with 1000-way parallelism. Our use of powerful general software subsystems will enable many other applications that need integration of streaming and batch data analytics.

Understanding the dynamics behind group formation and evolution in social networks is considered an instrumental milestone to better describe how individuals gather and form communities, how they enjoy and share the platform contents, how they are driven by their preferences/tastes, and how their behaviors are influenced by peers. In this context, the notion of compactness of a social group is particularly relevant. While the literature usually refers to compactness as a measure to merely determine how much members of a group are similar among each other, we argue that the mutual trustworthiness between the members should be considered as an important factor in defining such a term. In fact, trust has profound effects on the dynamics of group formation and their evolution: individuals are more likely to join with and stay in a group if they can trust other group members. In this paper, we propose a quantitative measure of group compactness that takes into account both the similarity and the trustworthiness among users, and we present an algorithm to optimize such a measure. We provide empirical results, obtained from the real social networks EPINIONS and CIAO, that compare our notion of compactness versus the traditional notion of user similarity, clearly proving the advantages of our approach.

Modern online social platforms allow their members to be involved in a broad range of activities including getting friends, joining groups, posting, and commenting resources. In this paper, we investigate whether a correlation emerges across the different activities a user can take part in. For our analysis, we focused on aNobii, a social platform with a world-wide user base of book readers, who post their readings, give ratings, review books, and discuss them with friends and fellow readers. aNobii presents a heterogeneous structure: 1) part social network, with user-to-user interactions; 2) part interest network, with the management of book collections; and 3) part folksonomy, with books that are tagged by the users. We analyzed a complete snapshot of aNobii and we focused on three specific activities a user can perform, namely tagging behavior, tendency to join groups and aptitude to compile a wishlist of the books one is planning to read. For each user, we create a tag-based, a group-based, and a wishlist-based profile. Experimental analysis, which was carried out with information-theory tools like entropy and mutual information, suggests that tag-based and group-based profiles are in general more informative than wishlist-based ones. Furthermore, we discover that the degree of correlation between the three profiles associated with the same user tend to be small. Hence, user profiling cannot be reduced to considering just any one type of user activity (albeit important) but it is crucial to incorporate multiple dimensions to effectively describe users' preferences and behavior.

The increasing pervasiveness of social media creates new opportunities to study human social behavior, while challenging our capability to analyze their massive data streams. One of the emerging tasks is to distinguish between different kinds of activities, for example engineered misinformation campaigns versus spontaneous communication. Such detection problems require a formal definition of meme, or unit of information that can spread from person to person through the social network. Once a meme is identified, supervised learning methods can be applied to classify different types of communication. The appropriate granularity of a meme, however, is hardly captured from existing entities such as tags and keywords. Here we present a framework for the novel task of detecting memes by clustering messages from large streams of social data. We evaluate various similarity measures that leverage content, metadata, network features, and their combinations. We also explore the idea of pre-clustering on the basis of existing entities. A systematic evaluation is carried out using a manually curated dataset as ground truth. Our analysis shows that pre-clustering and a combination of heterogeneous features yield the best trade-off between number of clusters and their quality, demonstrating that a simple combination based on pairwise maximization of similarity is as effective as a non-trivial optimization of parameters. Our approach is fully automatic, unsupervised, and scalable for real-time detection of memes in streaming data.

In this article, we propose a measure to assess scientific impact that discounts self‐citations and does not require any prior knowledge of their distribution among publications. This index can be applied to both researchers and journals. In particular, we show that it fills the gap of the h‐index and similar measures that do not take into account the effect of self‐citations for authors or journals impact evaluation. We provide 2 real‐world examples: First, we evaluate the research impact of the most productive scholars in computer science (according to DBLP Computer Science Bibliography, Universität Trier, Trier, Germany); then we revisit the impact of the journals ranked in the Computer Science Applications section of the SCImago Journal & Country Rank ranking service (Consejo Superior de Investigaciones Científicas, University of Granada, Extremadura, Madrid, Spain). We observe how self‐citations, in many cases, affect the rankings obtained according to different measures (including h‐index and ch‐index), and show how the proposed measure mitigates this effect.

In this work we present an in-depth analysis of the user behaviors on different Social Sharing systems. We consider three popular platforms, Flickr, Delicious and StumbleUpon, and, by combining techniques from social network analysis with techniques from semantic analysis, we characterize the tagging behavior as well as the tendency to create friendship relationships of the users of these platforms. The aim of our investigation is to see if (and how) the features and goals of a given Social Sharing system reflect on the behavior of its users and, moreover, if there exists a correlation between the social and tagging behavior of the users. We report our findings in terms of the characteristics of user profiles according to three different dimensions: (i) intensity of user activities, (ii) tag-based characteristics of user profiles, and (iii) semantic characteristics of user profiles.

Social (or folksonomic) tagging has become a very popular way to describe content within Web 2.0 websites. However, as tags are informally defined, continually changing, and ungoverned, it has often been criticised for lowering, rather than increasing, the efficiency of searching. To address this issue, a variety of approaches have been proposed that recommend users what tags to use, both when labeling and when looking for resources. These techniques work well in dense folksonomies, but they fail to do so when tag usage exhibits a power law distribution, as it often happens in real-life folksonomies. To tackle this issue, we propose an approach that induces the creation of a dense folksonomy, in a fully automatic and transparent way: when users label resources, an innovative tag similarity metric is deployed, so to enrich the chosen tag set with related tags already present in the folksonomy. The proposed metric, which represents the core of our approach, is based on the mutual reinforcement principle. Our experimental evaluation proves that the accuracy and coverage of searches guaranteed by our metric are higher than those achieved by applying classical metrics.

In this paper we present a novel strategy to discover the community structure of (possibly, large) networks. This approach is based on the well-know concept of network modularity optimization. To do so, our algorithm exploits a novel measure of edge centrality, based on the κ-paths. This technique allows to efficiently compute a edge ranking in large networks in near linear time. Once the centrality ranking is calculated, the algorithm computes the pairwise proximity between nodes of the network. Finally, it discovers the community structure adopting a strategy inspired by the well-known state-of-the-art Louvain method (henceforth, LM), efficiently maximizing the network modularity. The experiments we carried out show that our algorithm outperforms other techniques and slightly improves results of the original LM, providing reliable results. Another advantage is that its adoption is naturally extended even to unweighted networks, differently with respect to the LM.

Matrix Factorization techniques have been successfully applied to raise the quality of suggestions generated by Collaborative Filtering Systems (CFSs). Traditional CFSs based on Matrix Factorization operate on the ratings provided by users and have been recently extended to incorporate demographic aspects such as age and gender. In this paper we propose to merge CFS based on Matrix Factorization and information regarding social friendships in order to provide users with more accurate suggestions and rankings on items of their interest. The proposed approach has been evaluated on a real-life online social network; the experimental results show an improvement against existing CFSs. A detailed comparison with related literature is also present.