Search

In this work, we analyze what effect streaming gameplay on Twitch has on players’ in-game behavior and performance. We hypothesized that streaming can act as a form of implicit incentive to boost players’ performance and engagement. To test this hypothesis, we continuously collected data about all Twitch streams related to a popular Multiplayer Online Battle Arena (MOBA) game, League of Legends (LoL), and data of all LoL matches played during the same time frame, and cross-mapped the two data sets. We found that, counterintuitively, streaming significantly deteriorates players’ in-game performance: This may be due to the burden of carrying out two cognitively intensive activities at the same time, namely, playing the game and producing its commentary for streaming purposes. On the other hand, streaming increases engagement keeping players in significantly longer game sessions. We investigate these two effects further, to characterize how they vary upon individual characteristics.

Cooperation is a fundamental social mechanism, whose effects on human performance have been investigated in several environments. Online games are modern-days natural settings in which cooperation strongly affects human behavior. Every day, millions of players connect and play together in team-based games: the patterns of cooperation can either foster or hinder individual skill learning and performance. This work has three goals: (i) identifying teammates' influence on players' performance in the short and long term, (ii) designing a computational framework to recommend teammates to improve players' performance, and (iii) setting to demonstrate that such improvements can be predicted via deep learning. We leverage a large dataset from Dota 2, a popular Multiplayer Online Battle Arena game. We generate a directed co-play network, whose links' weights depict the effect of teammates on players' performance. Specifically, we propose a measure of network influence that captures skill transfer from player to player over time. We then use such framing to design a recommendation system to suggest new teammates based on a modified deep neural autoencoder and we demonstrate its state-of-the-art recommendation performance. We finally provide insights into skill transfer effects: our experimental results demonstrate that such dynamics can be predicted using deep neural networks.

Simulating and predicting planetary-scale techno-social systems poses heavy computational and modeling challenges. The DARPA SocialSim program set the challenge to model the evolution of GitHub, a large collaborative software-development ecosystem, using massive multi-agent simulations. We describe our best performing models and our agent-based simulation framework, which we are currently extending to allow simulating other planetary-scale techno-social systems. The challenge problem measured participant's ability, given 30 months of meta-data on user activity on GitHub, to predict the next months' activity as measured by a broad range of metrics applied to ground truth, using agent-based simulation. The challenge required scaling to a simulation of roughly 3 million agents producing a combined 30 million actions, acting on 6 million repositories with commodity hardware. It was also important to use the data optimally to predict the agent's next moves. We describe the agent framework and the data analysis employed by one of the winning teams in the challenge. Six different agent models were tested based on a variety of machine learning and statistical methods. While no single method proved the most accurate on every metric, the broadly most successful sampled from a stationary probability distribution of actions and repositories for each agent. Two reasons for the success of these agents were their use of a distinct characterization of each agent, and that GitHub users change their behavior relatively slowly.

How is popularity gained online? Is being successful strictly related to rapidly becoming viral in an online platform, or is it possible to acquire popularity in a steady and disciplined fashion? What are other temporal characteristics that can unveil the popularity of online content? To answer these questions, we leverage a multifaceted temporal analysis of the evolution of popular online content. We present dipm-SC: a multidimensional shape-based time-series clustering algorithm with a heuristic to find the optimal number of clusters. First, we validate the accuracy of our algorithm on synthetic datasets generated from benchmark time series models. Second, we show that dipm-SC can uncover meaningful clusters of popularity behaviors in real-world GitHub and Twitter datasets. By clustering the multidimensional time-series of the popularity of contents coupled with other domain-specific dimensions, we discover two main patterns of popularity: bursty and steady temporal behaviors. Furthermore, we find that the way popularity is gained over time has no significant impact on the final cumulative popularity.

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.

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 studying 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 dataset 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.

Online financial markets can be represented as complex systems where trading dynamics can be captured and characterized at different resolutions and time scales. In this work, we develop a methodology based on non-negative tensor factorization (NTF) aimed at extracting and revealing the multi-timescale trading dynamics governing online financial systems. We demonstrate the advantage of our strategy first using synthetic data, and then on real-world data capturing all interbank transactions (over a million) occurred in an Italian online financial market (e-MID) between 2001 and 2015. Our results demonstrate how NTF can uncover hidden activity patterns that characterize groups of banks exhibiting different trading strategies (normal vs. early vs. flash trading, etc.). We further illustrate how our methodology can reveal "crisis modalities" in trading triggered by endogenous and exogenous system shocks: as an example, we reveal and characterize trading anomalies in the midst of the 2008 financial crisis., Comment: 16 pages, 9 figures

The effects of team collaboration on performance have been explored in a variety of settings. Online games enable people with significantly different skills to cooperate and compete within a shared context. Players can affect teammates' performance either via direct communication or by influencing teammates' actions. Understanding such effects can help us provide insights into human behavior as well as make team recommendations. In this work, we aim at recommending teammates to each individual player for maximal skill growth. We study the effect of collaboration in online games using a large dataset from Dota 2, a popular Multiplayer Online Battle Arena game. To this aim, we construct an online co-play teammate network of players, whose links are weighted based on the gain in skill achieved due to team collaboration. We then use the performance network to devise a recommendation system based on a modified deep neural network autoencoder method.

Complex real-world challenges are often solved through teamwork. Of special interest are ad hoc teams assembled to complete some task. Many popular multiplayer online battle arena (MOBA) video-games adopt this team formation strategy and thus provide a natural environment to study ad hoc teams. Our work examines data from a popular MOBA game, League of Legends, to understand the evolution of individual performance within ad hoc teams. Our analysis of player performance in successive matches of a gaming session demonstrates that a player’s success deteriorates over the course of the session, but this effect is mitigated by the player’s experience. We also find no significant long-term improvement in the individual performance of most players. Modelling the short-term performance dynamics allows us to accurately predict when players choose to continue to play or end the session. Our findings suggest possible directions for individualized incentives aimed at steering the player’s behaviour and improving team performance.

We introduce a system for automatically generating warnings of imminent or current cyber-threats. Our system leverages the communication of malicious actors on the darkweb, as well as activity of cyber security experts on social media platforms like Twitter. In a time period between September, 2016 and January, 2017, our method generated 661 alerts of which about 84% were relevant to current or imminent cyber-threats. In the paper, we first illustrate the rationale and workflow of our system, then we measure its performance. Our analysis is enriched by two case studies: the first shows how the method could predict DDoS attacks, and how it would have allowed organizations to prepare for the Mirai attacks that caused widespread disruption in October 2016. Second, we discuss the method's timely identification of various instances of data breaches.

Widespread adoption of networking technologies has brought about tremendous economic and social growth, but also exposed individuals and organization to new threats from malicious cyber actors. Recent attacks by WannaCry and NotPetya ransomware crypto-worms, infected hundreds of thousands of computer systems world wide, compromising data and critical infrastructure. In order to limit their impact, it is, therefore, critical to detect---and even predict---cyber attacks before they spread. Here, we introduce DISCOVER, an early cyber threat warning system, that mines online chatter from cyber actors on social media, security blogs, and darkweb forums, to identify words that signal potential cyber attacks. We evaluate DISCOVER and find that it can identify terms related to emerging cyber threats with precision above $80%$. DISCOVER also generates a time line of related online discussions on different Web sources that can be useful for analyzing emerging cyber threats.

Recent advances in data collection have facilitated the access to time-resolved human proximity data that can conveniently be represented as temporal networks of contacts between individuals. While the structural and dynamical information revealed by this type of data is fundamental to investigate how information or diseases propagate in a population, data often suffer from incompleteness, which possibly leads to biased estimations in data-driven models. A major challenge is thus to estimate the outcome of spreading processes occurring on temporal networks built from partial information. To cope with this problem, we devise an approach based on non-negative tensor factorization, a dimensionality reduction technique from multilinear algebra. The key idea is to learn a low-dimensional representation of the temporal network built from partial information and to use it to construct a surrogate network similar to the complete original network. To test our method, we consider several human-proximity networks, on which we perform resampling experiments to simulate a loss of data. Using our approach on the resulting partial networks, we build a surrogate version of the complete network for each. We then compare the outcome of a spreading process on the complete networks (nonaltered by a loss of data) and on the surrogate networks. We observe that the epidemic sizes obtained using the surrogate networks are in good agreement with those measured on the complete networks. Finally, we propose an extension of our framework that can leverage additional data, when available, to improve the surrogate network when the data loss is particularly large.

Background: Over the past several decades, naturally occurring and man-made mass casualty incidents (MCIs) have increased in frequency and number worldwide. To test the impact of such events on medical resources, simulations can provide a safe, controlled setting while replicating the chaotic environment typical of an actual disaster. A standardized method to collect and analyze data from mass casualty exercises is needed to assess preparedness and performance of the health care staff involved. Objective: In this study, we aimed to assess the feasibility of using wearable proximity sensors to measure proximity events during an MCI simulation. In the first instance, our objective was to demonstrate how proximity sensors can collect spatial and temporal information about the interactions between medical staff and patients during an MCI exercise in a quasi-autonomous way. In addition, we assessed how the deployment of this technology could help improve future simulations by analyzing the flow of patients in the hospital. Methods: Data were obtained and collected through the deployment of wearable proximity sensors during an MCI functional exercise. The scenario included 2 areas: the accident site and the Advanced Medical Post, and the exercise lasted 3 hours. A total of 238 participants were involved in the exercise and classified in categories according to their role: 14 medical doctors, 16 nurses, 134 victims, 47 Emergency Medical Services staff members, and 27 health care assistants and other hospital support staff. Each victim was assigned a score related to the severity of his/her injury. Each participant wore a proximity sensor, and in addition, 30 fixed devices were placed in the field hospital. Results: The contact networks show a heterogeneous distribution of the cumulative time spent in proximity by the participants. We obtained contact matrices based on the cumulative time spent in proximity between the victims and rescuers. Our results showed that the time spent in proximity by the health care teams with the victims is related to the severity of the patient’s injury. The analysis of patients’ flow showed that the presence of patients in the rooms of the hospital is consistent with the triage code and diagnosis, and no obvious bottlenecks were found. Conclusions: Our study shows the feasibility of the use of wearable sensors for tracking close contacts among individuals during an MCI simulation. It represents, to our knowledge, the first example of unsupervised data collection—ie, without the need for the involvement of observers, which could compromise the realism of the exercise—of face-to-face contacts during an MCI exercise. Moreover, by permitting detailed data collection about the simulation, such as data related to the flow of patients in the hospital, such deployment provides highly relevant input for the improvement of MCI resource allocation and management.

Mobilization of neutrophils from the bone marrow determines neutrophil blood counts and thus is medically important. Balanced neutrophil mobilization from the bone marrow depends on the retention-promoting chemokine CXCL12 and its receptor CXCR4 and the egression-promoting chemokine CXCL2 and its receptor CXCR2. Both pathways activate the small guanosine triphosphatase Rac, leaving the role of this signaling event in neutrophil retention and egression ambiguous. On the assumption that active Rac determines persistent directional cell migration, we generated a mathematical model to link chemokine-mediated Rac modulation to neutrophil egression time. Our computer simulation indicated that, in the bone marrow, where the retention signal predominated, egression time strictly depended on the time it took Rac to return to its basal activity (namely, adaptation). This prediction was validated in mice lacking the Rac inhibitor ArhGAP15. Neutrophils in these mice showed prolonged Rac adaptation and cell-autonomous retention in the bone marrow. Our model thus demonstrates that mobilization in the presence of two spatially defined opposing chemotactic cues strictly depends on inhibitors shaping the time course of signal adaptation. Furthermore, our findings might help to find new modes of intervention to treat conditions characterized by excessively low or high circulating neutrophils.

Multiplayer online battle arena has become a popular game genre. It also received increasing attention from our research community because they provide a wealth of information about human interactions and behaviors. A major problem is extracting meaningful patterns of activity from this type of data, in a way that is also easy to interpret. Here, we propose to exploit tensor decomposition techniques, and in particular Non-negative Tensor Factorization, to discover hidden correlated behavioral patterns of play in a popular game: League of Legends. We first collect the entire gaming history of a group of about one thousand players, totaling roughly $100K$ matches. By applying our methodological framework, we then separate players into groups that exhibit similar features and playing strategies, as well as similar temporal trajectories, i.e., behavioral progressions over the course of their gaming history: this will allow us to investigate how players learn and improve their skills., Comment: 9 pages, 6 figures, submitted to KDD'17

For better and worse, our world has been transformed by Big Data. To understand digital traces generated by individuals, we need to design multidisciplinary approaches that combine social and data science. Data and social scientists face the challenge of effectively building upon each other’s approaches to overcome the limitations inherent in each side. Here, we offer a “data science perspective” on the challenges that arise when working to establish this interdisciplinary environment. We discuss how we perceive the differences and commonalities of the questions we ask to understand digital behaviors (including how we answer them), and how our methods may complement each other. Finally, we describe what a path toward common ground between these fields looks like when viewed from data science.