Search for: All records

Experiments that are games played among a network of players are widely used to study human behavior. Furthermore, bots or intelligent systems can be used in these games to produce contexts that elicit particular types of human responses. Bot behaviors could be specified solely based on experimental data. In this work, we take a different perspective, called the Probability Calibration (PC) approach, to simulate networked group anagram games with certain players having bot-like behaviors. The proposed method starts with data-driven models and calibrates in principled ways the parameters that alter player behaviors. It can alter the performance of each type of agent (e.g., bot) in group anagram games. Further, statistical methods are used to test whether the PC models produce results that are statistically different from those of the original models. Case studies demonstrate the merits of the proposed method. 

Common knowledge (CK) is a phenomenon where each individual within a group knows the same information and everyone knows that everyone knows the information, infinitely recursively. CK spreads information as a contagion through social networks in ways different from other models like susceptible-infectious-recovered (SIR) model. In a model of CK on Facebook, the biclique serves as the characterizing graph substructure for generating CK, as all nodes within a biclique share CK through their walls. To understand the effects of network structure on CK-based contagion, it is necessary to control the numbers and sizes of bicliques in networks. Thus, learning how to generate these CK networks (CKNs) is important. Consequently, we develop an exponential random graph model (ERGM) that constructs networks while controlling for bicliques. Our method offers powerful prediction and inference, reduces computational costs significantly, and has proven its merit in contagion dynamics through numerical experiments. 

When modeling human behavior in multi-player games, it is important to understand heterogeneous aspects of player behaviors. By leveraging experimental data and agent-based simulations, various data-driven modeling methods can be applied. This provides a great opportunity to quantify and visualize the uncertainty associated with these methods, allowing for a more comprehensive understanding of the individual and collective behaviors among players. For networked anagram games, player behaviors can be heterogeneous in terms of the number of words formed and the amount of cooperation among networked neighbors. Based on game data, these games can be modeled as discrete dynamical systems characterized by probabilistic state transitions. In this work, we present both Frequentist and Bayesian approaches for visualizing uncertainty in networked anagram games. These approaches help to elaborate how players individually and collectively form words by sharing letters with their neighbors in a network. Both approaches provide valuable insights into inferring the worst, average, and best player performance within and between behavioral clusters. Moreover, interesting contrasts between the Frequentist and Bayesian approaches can be observed. The knowledge and inferences gained from these approaches are incorporated into an agent-based simulation framework to further demonstrate model uncertainty and players’ heterogeneous behaviors. 

In group anagram games, players cooperate to form words by sharing letters that they are initially given. The aim is to form as many words as possible as a group, within five minutes. Players take several different actions: requesting letters from their neighbors, replying to letter requests, and forming words. Agent-based models (ABMs) for the game compute likelihoods of each player’s next action, which contain uncertainty, as they are estimated from experimental data. We adopt a Bayesian approach as a natural means of quantifying uncertainty, to enhance the ABM for the group anagram game. Specifically, a Bayesian nonparametric clustering method is used to group player behaviors into different clusters without pre-specifying the number of clusters. Bayesian multi nominal regression is adopted to model the transition probabilities among different actions of the players in the ABM. We describe the methodology and the benefits of it, and perform agent-based simulations of the game. 

There are myriad real-life examples of contagion processes on human social networks, e.g., spread of viruses, information, and social unrest. Also, there are many methods to control or block contagion spread. In this work, we introduce a novel method of blocking contagions that uses nodes from dominating sets (DSs). To our knowledge, this is the first use of DS nodes to block contagions. Finding minimum dominating sets of graphs is an NP-Complete problem, so we generalize a well-known heuristic, enabling us to customize its execution. Our method produces a prioritized list of dominating nodes, which is, in turn, a prioritized list of blocking nodes. Thus, for a given network, we compute this list of blocking nodes and we use it to block contagions for all blocking node budgets, contagion seed sets, and parameter values of the contagion model. We report on computational experiments of the blocking efficacy of our approach using two mined networks. We also demonstrate the effectiveness of our approach by comparing blocking results with those from the high degree heuristic, which is a common standard in blocking studies. 

Heterogeneous player behaviors are commonly observed in games. It is important to quantify and visualize these heterogeneities in order to understand collective behaviors. Our work focuses on developing a Bayesian approach for uncertainty visualization in a model of networked anagram games. In these games, team members collectively form as many words as possible by sharing letters with their neighbors in a network. Heterogeneous player behaviors include great differences in numbers of words formed and the amount of cooperation among networked neighbors. Our Bayesian approach provides meaningful insights for inferring worst, average, and best player performance within behavioral clusters, overcoming previous model shortcomings. These inferences are integrated into a simulation framework to understand the implications of model uncertainty and players' heterogeneous behaviors. 

Motivated by a wide range of applications, research on agent-based models of contagion propagation over networks has attracted a lot of attention in the literature. Many of the available software systems for simulating such agent-based models require users to download software, build the executable, and set up execution environments. Further, running the resulting executable may require access to high performance computing clusters. Our work describes an open access software system (NetSimS) that works under the “Modeling and Simulation as a Service” (MSaaS) paradigm. It enables users to run simulations by selecting models and parameter values, initial conditions, and networks through a web interface. The system supports a variety of models and networks with millions of nodes and edges. In addition to the simulator, the system includes components that enable users to choose initial conditions for simulations in a variety of ways, to analyze the data generated through simulations, and to produce plots from the data. We describe the components of NetSimS and carry out a performance evaluation of the system. We also discuss two case studies carried out on large networks using the system. NetSimS is a major component within net.science, a cyberinfrastructure for network science. 

Data from surveys administered after Hurricane Sandy provide a wealth of information that can be used to develop models of evacuation decision-making. We use a model based on survey data for predicting whether or not a family will evacuate. The model uses 26 features for each household including its neighborhood characteristics. We augment a 1.7 million node household-level synthetic social network of Miami, Florida with public data for the requisite model features so that our population is consistent with the survey-based model. Results show that household features that drive hurricane evacuations dominate the effects of specifying large numbers of families as \early evacuators" in a contagion process, and also dominate effects of peer influence to evacuate. There is a strong network-based evacuation suppression effect from the fear of looting. We also study spatial factors affecting evacuation rates as well as policy interventions to encourage evacuation. 

We study evacuation dynamics in a major urban region (Miami, FL) using a combination of a realistic population and social contact network, and an agent-based model of evacuation behavior that takes into account peer influence and concerns of looting. These factors have been shown to be important in prior work, and have been modeled as a threshold-based network dynamical systems model (2mode-threshold), which involves two threshold parameters|for a family's decision to evacuate and to remain in place for looting and crime concerns|based on the fraction of neighbors who have evacuated. The dynamics of such models are not well understood, and we observe that the threshold parameters have a significant impact on the evacuation dynamics. We also observe counter-intuitive effects of increasing the evacuation threshold on the evacuated fraction in some regimes of the model parameter space, which suggests that the details of realistic networks matter in designing policies. 

Neighborhood effects have an important role in evacuation decision-making by a family. Owing to peer influence, neighbors evacuating can motivate a family to evacuate. Paradoxically, if a lot of neighbors evacuate, then the likelihood of an individual or family deciding to evacuate decreases, for fear of crime and looting. Such behavior cannot be captured using standard models of contagion spread on networks, e.g., threshold, independent cascade, and linear threshold models. Here, we propose a new threshold-based graph dynamical system model, 2mode-threshold, which captures this dichotomy. We study theoretically the dynamical properties of 2mode-threshold in different networks, and find significant differences from a standard threshold model. We build and characterize small world networks of Virginia Beach, VA, where nodes are geolocated families (households) in the city and edges are interactions between pairs of families. We demonstrate the utility of our behavioral model through agent-based simulations on these small world networks. We use it to understand evacuation rates in this region, and to evaluate the effects of modeling parameters on evacuation decision dynamics. Specifically, we quantify the effects of (1) network generation parameters, (2) stochasticity in the social network generation process, (3) model types (2mode-threshold vs. standard threshold models), (4) 2mode-threshold model parameters, (5) and initial conditions, on computed evacuation rates and their variability. An illustrative example result shows that the absence of looting effect can overpredict evacuation rates by as much as 50%.