1,721,063 research outputs found
Recommended from our members
Spectral Analysis: a New Perspective for Mining Human Activity Data
Get PDFThe pervasive usage of mobile phones provides an abundance of data associated with their users. Data generated by mobile devices includes measurements of user actions (e.g. call logs data, SMS logs data, app usage sessions data) as well as records regarding its local environment (e.g. data about other local blue tooth devices, GPS/location trajectories). These data provide information regarding the underlying behaviors of device users and their corresponding social contexts, thus giving us new opportunities for learning human spatio-temporal and social patterns.In this dissertation, new methods are developed to mine data generated by mobile devices and further improve applications in the area of social learning and inference. Particularly, three objectives are pursued here: (1) using mobile phone data to distinguish dyads embedded in social relationships from those that are not; (2) understanding how human telecommunication activities are associated with urban ecology; and (3) obtaining insights regarding users' demographic features from mobile app usage data. A key characteristic of these approaches is that they leverage activity and environmental data at the device level (in some cases, aggregated across sets of users), often allowing the data in question to be collected and stored in an efficient and privacy preserving manner. In pursuing the first objective, we introduce a new approach --\emph{activity correlation spectroscopy} -- to inferring relationships by exploiting the spectral and distributional structure of activity correlation within dyads. Unlike existing techniques, our approach can be employed with minimal, individual-level (i.e., non-relational), and non-identifying data that is easily collected using commodity hardware. We demonstrate our methodology via an application to detection of friendship and group co-membership using mobile device and survey data from the MIT Reality Mining study \citep{eagle2009}.Vis a vis the second objective, we provide a novel approach to the use of spatio-temporally aggregated cell phone data to learn features of urban ecology (i.e., spatial distributions of distinct social and economic entities and their associated activities). Specifically, our technique involves four stages: (\RN{1}) decomposing the aggregated cell phone activity within local areal units using spectral methods; (\RN{2}) learning spectral characteristics associated with ecological features using a training set; (\RN{3}) predicting local ecology composition for out-of-sample areas; and (\RN{4}) predicting activity time series for out-of-sample areas. The core of our approach is the projection of spectral features in cell phone activity series into an ecology-associated basis, allowing both identification of communication patterns arising from particular types of local activities and/or institutions and leveraging of those patterns for classification and activity prediction. We apply our methodology to aggregated communication and Internet traffic data from the cities of Milan and Trento to show the effectiveness of our method. Finally, in pursuing the third objective, we demonstrate an integrated system which can cheaply and easily collect application behavior and survey data from mobile phones; we introduce several novel features that assist the learning of individual level demographic features (e.g., gender and age group). Specifically, our approach for learning and inference for demographic features involves new techniques: (\RN{1}) decomposing the app usage from mobile phones using spectral methods; (\RN{2}) learning spectral characteristics associated with individuals using a training set; (\RN{3}) combining other temporal features with learned spectral characteristics to predict demographic features for out-of-sample individuals. The core of our methodology is the utilization of spectral features in cell phone app activity series, allowing both identification of behavior patterns arising from particular types of cell phone apps and leveraging of those patterns for demographic classification and prediction. We demonstrate the effectiveness of our approach with an application to real mobile app traffic data from the United States
Recommended from our members
The Geography and Multiplexity of Personal Networks
Get PDFWhile the importance of geographic space in structuring social interactions is well established, much has remained unknown to researchers due in large part to the lack of available data on personal networks across space. Utilizing data from the American Social Fabric Study (Butts et al., 2014), a spatially stratified egocentric network study, this thesis aims to answer various methodological and substantive questions relating to networks and geography that have not been studied previously. The first paper is a methodological piece exploring what factors are related to the precision of geographic location of survey respondents (egos) as well as their social ties on six relations (alters). The second study examines an oft-studied social relation, that of the job seeking social tie, from a new perspective. In particular, we focus on the pool of potential job lead ties (rather than those that were mobilized) and study the relation in terms of multiplexity (i.e., the overlap of social relations) and locality to ego. The final paper examines another vital social relation: the ties that ego would seek to notify in the event of a disaster or emergency affecting his or her area. Given that this type of phenomenon is inherently spatial, we seek to understand the extent to which these ties are both spatially and socially embedded
Recommended from our members
Frontier Problems in Modeling Dynamic Social Systems
Get PDFSocial dynamics – changes in the structure, existence or occurrence of social connections (e.g. sexual, friendship) or events (e.g. conversation, armed conflict) – are a core interest of social sciences (Comte, 1855; Weber, 1904; Marx and Engels, 1972). One key to understanding dynamic social systems is to understand the social networks that comprise these systems (Mayhew, 1984; Moreno, 1934). Modeling networks over time is of essential interest to understanding social systems and structure, from intimate sexual partnership ties (Lévi-Strauss, 1969; Morris and Kretzschmar, 1995; Hudson, 1993; Moody, 2002) to the interactions of large-scale organizations (Blau, 1970; Childers et al., 1971; Mayhew et al., 1972). Recently, advances in computing and statistical methodology have improved our ability to model relational dynamics. One such methodology is the exponential family models for modeling social networks (i.e. temporal exponential random graph models (TERGMs), see (Krivitsky and Handcock, 2014)). This thesis aims to improve our ability to model relational dynamics through a substantive exploration of a dynamic social system, a purely predictive imputation of edge dynamics among social media accounts, and an exploration of how a dynamic model handles temporal adjustments. I thus aim to address frontier problems in relational dynamics, to improve and extend our current models by both substantive application and methodology assessment
Recommended from our members
Using Network Models to Relate Local Interactions with Global Topology: Applications to Protein Interactions and Emergent Multi-Body Structures
Get PDFLocal interactions within and between proteins (or interacting objects in general) inherently determine the resulting global structure, whether that be a monomeric protein structure, a dimer or multimer, or a larger aggregate consisting of tens to thousands of proteins. For proteins, structure is canonically partitioned into four levels: primary, which describes the sequence of residues that make up the protein; secondary, the α-helices and β-sheets that result from hydrogen-bonding interactions between residues; tertiary, which describes (somewhat arbitrarily defined) domains of clustered secondary structures that are typically held together with salt-bridges; and finally, quaternary structures composed of multiple proteins interacting via hydrogen-bonding or other polar interactions. Variants are proteins with point mutations, or mutations occurring to a small number (typically one) of the amino acids in the primary structure. Point mutations can alter the higher-order structure and dynamics of the protein, and thus how it responds to its environment, making it susceptible to evolutionary forces that dampen or put emphasis on a given variant. Such changes in structure and dynamics can range from subtle deformations to changes in the way the protein folds, inhibiting function. Mutations that are favored by evolution provide information about how the protein’s relationship with its environment affects its function and applies pressure to the adaptative evolution of the protein. The effects of mutations on protein structure, function, and interactions are explored in chapters two and three of this text. To contrast, the fourth chapter takes a generalized approach by delving into the range of emergent multi-body structures that can arise from slight changes in environmental or structural parameters while remaining agnostic to any specific features of a single protein sequence
Recommended from our members
Large-Scale Analyses of Organizational Disaster and Hazard Communications
Get PDFThis dissertation analyzes various social behaviors in disaster and hazard contexts by analyzing and extracting data from large-scale organizational textual communications. The chapters in this dissertation present broad applications of social network analysis, natural language processing, mathematical models, along with theories of social science to address several facets of individual and organizational interactions within a hazards context. Chapter 2 examines the relationships between different message-dependent actions by building a simple, first-order mathematical model that infers on correlations between the underlying drivers from aggregated data. This chapter also includes an empirical application of the framework to the case of hazards messaging, finding that transmission-driving factors are distinct from those driving engagement. Chapter 3 analyzes the semantic network of causal narratives employed by officials during the COVID-19 pandemic, a reflection of the broader discourse built by a large set of entities. This chapter examines the structure of the extracted causal narrative network, the dynamics of narrative usage across various actors and across time, and how concept positions within the structure of causal discourse impact message retransmission. Chapter 4 examines how organizational task performance affects collaborations within an emergent multiorganizational network during a large-scale disaster response. This chapter builds a large language model-based workflow for measuring task performance from thousands of long-form reports detailing organizational activities during the 2005 Hurricane Katrina response. Subsequent analysis with extracted task structures describes the tasks performed by thousands of entities, revealing nuances in the formation of collaborative ties during Hurricane Katrina. Chapter 5 is an overview on the series of experiments undertaken in improving the accuracy of large language models for extracting data from hazards-related texts. Although most experiments were unsuccessful, the few successes could prove promising for future research efforts in coding massive corpuses of domain-specific texts
Recommended from our members
Advances in Exponential-family Random Graph Models: Computation, Model Selection, and Methodology
Get PDFNetworks (graphs) are broadly used to represent relations between entities in a wide range of scientific fields. Exponential-family random graph models (ERGMs) provide a highly general way of specifying distributions on graphs, allowing the complex dependence structure of edges in a network to be specified in terms of local structural properties. This thesis addresses problems related to three lines of inquiry for ERGMs: faster Bayesian inference algorithms; comparison of newly proposed and traditional model selection techniques; and methodological innovation for modeling ensembles of networks. In Chapter 2 of this dissertation, we present a highly parallel algorithm that enables fast Bayesian inference on ERGMs. The impetus for this work comes from the facts that conducting Bayesian inference for ERGMs is challenging because of the intractability of both the likelihood and posterior normalizing factor and auxiliary-variable based Markov Chain Monte Carlo (MCMC) methods for this problem are asymptotically exact but computationally demanding. We propose a kernel-based approximate Bayesian computation algorithm for fitting ERGMs, which is easily parallelizable. Through empirical comparisons against the state-of-the-art approximate exchange algorithm, we show that the proposed algorithm yields comparable accuracy to the state-of-the-art MCMC approach, the approximate exchange algorithm (Caimo and Friel, 2011), while cutting the wallclock runtime by half with 5 cores, and by 80\% with 30 cores.In Chapter 3 of this dissertation, we carry out simulation studies to compare newly proposed and traditional model selection techniques. This work is driven by the importance of understanding the strengths and weaknesses of those model selection techniques for ERGMs that are currently available, including Akaike information criterion (Akaike, 1973), Bayesian information criterion (Schwarz, 1978), Held-Out Predictive Evaluation (HOPE) (Yin et al., 2019), Bayes factors (Raftery, 1995) and graphical goodness of fit (Hunter et al., 2008). In particular, we focus on the first three techniques, as the calculation of Bayes factor for ERGMs relies on reversible jump Markov chain Monte Carlo algorithm extension of the approximate exchange algorithm (Caimo and Friel, 2013), which is hard to implement and tune; the graphical goodness of fit is more suitable for checking whether a model is adequate rather than comparing competing models. The simulation studies are carried out under two scenarios, closed-M (under which the true model is among the set of candidate models) and open-M (under which the true model is not among the set of candidate models), and we evaluate the performance of model selection techniques from various aspects covering the model selection accuracy, predictive deviance and prediction accuracy of edge variables. In Chapter 4 of this dissertation, we propose a novel methodology that can be used for modeling the generative processes of ensembles of networks. The motivation of this work is that ensembles of networks arise in many scientific fields, but there are few statistical tools for inferring their generative processes, particularly in the presence of both dyadic dependence and cross-graph heterogeneity. To fill in this gap, we propose characterizing network ensembles via finite mixtures of exponential family random graph models, a framework for parametric statistical modeling of graphs that has been successful in explicitly modeling the complex stochastic processes that govern the structure of edges in a network. Our proposed methodology can also be used for applications such as model-based clustering of ensembles of networks and density estimation for complex graph distributions. We develop a Metropolis-within-Gibbs algorithm to conduct fully Bayesian inference and adapt a version of deviance information criterion for missing data models to choose the number of latent heterogeneous generative mechanisms. Simulation studies show that the proposed procedure can recover the true number of latent heterogeneous generative processes and corresponding parameters. We demonstrate the utility of the proposed approach using an ensemble of political co-voting networks among U.S. Senators and an ensemble of advice-seeking networks among school teachers
Recommended from our members
Communication Across the Spectrum of Hazards and Disasters
Get PDFThis dissertation investigates several instances of the micro-communication landscape across the spectrum of hazards, from the quotidian to the exotic, by offering a deeper understanding into the communication process via retransmission and communication dynamics. Chapter 2 focuses on hazard communication during quotidian and atypical hazards in the context of the National Weather Service's use of Twitter from 2009-2021. We investigate several micro-structural, content, and style related message features to understand the properties that make a message more likely to be retransmitted. Chapter 3 looks into communication occurring in the range of exotic and atypical end of the spectrum by studying public-health communicators on Twitter during the first eight months of the unfolding coronavirus disease 2019. Finally, Chapter 4 focuses solely on the exotic end of this spectrum in an investigation of 17 communication networks during the unfolding events of the 2001 World Trade Center Disaster. We model 17 dynamic radio networks to understand the role that the social mechanisms of preferential attachment, Institutionalized Coordinator Roles, and conversational inertia play in the communication process of a disrupted environment. This dissertation provides a holistic overview of hazard communication across this spectrum, providing into the kinds of micro-communication strategies and processes that are unfolding. We hope it inspires future research in this area of critical importance
Recommended from our members
In Case of Emergency, Don’t Break Glass: The Emergency Management Organizational Field as a Glassy Regime
Get PDFThe way that relations between organizations and their environments affect their structures important features of organizations. These relations can potentially induce conformity, and their environments can potentially maintain differences. Organizations dependent on a leading organization for resources are theorized to structure themselves more like the leading organization. Conversely, organizations that function in divergent social contexts are expected to structure themselves divergently. While social network analyses have offered insights into this domain, previous work typically compares organizational structures and not the mechanisms that generate them. The advent of exponential family random graph models enables the examination of the underlying mechanisms, or social forces, that produce networks such as organizational structures. However, the tools to evaluate the adequacy or assess the fit of these models have opened up new questions regarding model evaluation and adequacy assessment. The first chapter of this dissertation introduces several new methods for evaluating the adequacy of exponential random graph models using labeled rather than topological features. The second chapter advances a within-sample model validation algorithm to assess the fit of a model. These tools were used to develop exponential random graph models that are used in the third chapter, which analyzes the generative features of states’ emergency operations plans. I build a new dataset from primary source documents that delineate the organizations assigned to a standardized set of emergency support functions. I then evaluate the historical contingency of the realized emergency management networks within each state through simulation studies. These studies build upon previous work and facilitate a virtual "rerunning of history" to assess whether states form similar networks due to exogenous pressures from the federal government or whether the social substrate within each state drives observable differences. From these analyses, I postulate an interorganizational continuum that is analogous to the physical states of matter to explain the observed differences and similarities. As such, this dissertation makes several valuable contributions to both the network analytic methodology and theories of interorganizational relations
Recommended from our members
The Spacetime Continuum: Using Spatio-temporal Filtering Techniques to Revisit Rumor Theory
Get PDFA characteristic feature of disasters is the disruption of routine activities in a society. This disruption spurs novel activity from a variety of social units, from organizations to public officials to ordinary citizens. A major component of this response is informal communication among the public and this public communication increasingly occurs in online settings. Much of this online, informal communication falls under the formal definitions of rumoring and I harness the precision of Twitter's timestamped, geolocated messages to probe the spatial and temporal features of rumoring. A challenge of relying on streams of online data is finding signal amid the overwhelming volume of activity. To overcome this obstacle I use what is known about the spatial and temporal characteristics of both rumoring and disaster-related information transmission to develop a spatio-temporal filtering approach for measuring rumoring. I demonstrate in the first chapter that this approach offers strong increases in signal of hazard-related rumoring activity across a variety of events, both natural and anthropogenic. The results shed light onto the distribution of rumoring activity across large spatial scales. In the second chapter I use the same spatio-temporal filtering approach to identify surges in signal of content of hazard-related rumoring as I look for evidence of topical convergence and content evolution throughout the rumoring process. In the third chapter I test a variety of rumor theories by measuring which features of each U.S. county---its demographics, history of tornado events, and volume interpersonal ties to the tornado-affected county---determine its propensity to rumor about severe tornado events. Using data whose spatio-temporal precision and scope (both in geography and variety of events) have historically been infeasible, this dissertation makes several valuable contributions to rumor theories
Recommended from our members
Manufactured Diversity and Passing En Masse
Get PDFThis dissertation examines the mechanism behind state-sponsored disinformation campaigns, with a focus on the case of the Internet Research Agency's (IRA) use of a Social Influence Bot Network (SIBN) during 2016. Leaning on foundational sociological theories of passing and legitimation, this thesis examines how SIBN operators strategically balance competing needs for concealment and effective narrative spread. Through a mixed-methods approach that involves social network analysis, TF-IDF weighting, and case control design, this study uncovers the passing strategies employed by social bots to infiltrate online social spaces, the trade-offs SIBN operators must make to avoid detection while maintaining efficiency, and the dynamic adaptations of disinformation tactics in response to surrounding political contexts. Findings uncover the nine distinct persona archetypes the IRA's SIBN employed to infiltrate different online communities, the preference for efficiency over concealment in their link-sharing tactics, and the shifting strategy employed by the IRA in response to election milestones. This research contributes to the study of computational propaganda by providing a deep understanding of the operations involved in online actors manipulating public discourse
- …
