Facebook public pages are a popular form of online social network (OSN) communities. The "like" connections between public pages create a graph of pages on Facebook. Geographic location is a crucial piece of metadata for pages, but it is often omitted by page managers. We propose a classification algorithm to restore the missing subdivision location of Facebook public pages. We propose neighborhood state distribution vectors as features for graph neural networks to classify the state of the pages. Then, we define intrastate and interstate Facebook public pages based on the high-probability state label outputted by the classification model. Finally, we profile states with different influences over the online communities by analyzing the classification confusion matrix, interstate page percentages, and interstate pages across state borders. Our method achieves better accuracy (87.52%) and F1 score (0.8756) than previous studies (66.2% and 73.08%).

Online Social Network (OSN) communities serve as different platforms for multiple users' interaction - people behaving diversely among distinctive communities - such as entertainment, global and local discussion communities. However, attribute identification among online discussion communities remain largely unexplored. In this paper, we describe and analyze the geo-location property of large-scale Facebook public pages (15M pages). We propose a framework utilizing the connectivity of the page-like graph to predict the missing geo-location information based on Breadth-First Search (BFS). Our method achieves a satisfyingly high accuracy (89 %) on identifying the state location attribute of unknown United States (US) pages. Our empirical results offer a better understanding of regional social analysis and target audience broadcasting.

Mining for anomalies in graph structured datasets is an important and challenging problem for many applications including security, health care, and social media. In this paper, we propose a novel framework to localize temporal anomalies in large evolving graphs with reduced false alarm rate. Specifically, we first introduce a node-centric model based on Vector Autoregression to analyze node behavior history in dynamic graphs. Then we develop two community-centric models to reduce the amount of false positive results by tracking the structural change and dynamics of graph communities. We analyze the performance of our proposed anomaly localization framework on several synthetic and real-world data sets including Enron email network data, an enterprise network traffic data, and CNN public Facebook page. All experimental results show the effectiveness and consistency of our framework in localizing temporal anomalies with reduced false alarm rate.

With the popularity of social media in recent years, it has been a critical topic for social network designer to understand the factors that influence continued user participation in online newsgroups. Our study examines how feedback with different opinions is associated with participants' lifetime in online newsgroups. Firstly, we propose a new method of classifying different opinions among user interaction contents. Generally, we leverage user behavior information in online newsgroups to estimate their opinions and evaluate our classification results based on linguistic features. In addition, we also implement this opinion classification method into our SINCERE system as a real-time service. Based on this opinion classification tool, we use survival analysis to examine how others' feedback with different opinions influence continued participation. In our experiment, we analyze more than 88,770 interactions on the official Occupy LA Facebook page. Our final result shows that not only the feedback with the same opinions as the user, but also the feedback with different opinions can motivate continued user participation in online newsgroup. Furthermore, an interaction of feedback with both the same and different opinions can boost user continued participation to the greatest extent. This finding forms the basis of understanding how to improve online service in social media.

Social macrostructures, such as structural balance, ranked clusters and transitivity, are of great importance on account of their abilities to reflect the underlying social psychological processes about the formation and evolution of relationships among people. Here we present a detailed study on examining the existence and evolution of social macrostructures in an empirical online social network, and exploring how they can be explained by network micro structural properties, i.e. nodal in degree and out degree and dyadic feature. We establish the micro-macro linkage by analyzing the network triadic patterns. Based on a novel clustering coefficient based network sampling approach, we show that the distribution of observed triad census in our data is low dimensional and can be greatly explained by network dyadic properties. In a time series analysis, we observe that our network exhibits strong tendencies towards balanced, transitive and clustered social macrostructure given the nodal and dyadic characteristics. Our findings supplement the studies on structural properties of online social network by providing more insights on the relation between network macrostructures and the micro-level social processes that result in them. And they form the basis to understand better how online social media systems change the information and communication fabric of our society.

Online Social Network (OSN) operators are interested in promoting usage among their users, and try a variety of strategies to encourage use. Some recruit celebrities to their site, some allow third parties to develop applications that run on their sites, and all have features intended to encourage use. As important as usage is, we are unaware of any studies into what influences users to be active and to remain online. This paper is the first work studying the lifetime of OSN users, examining the factors that influence lifetime in one OSN, Buzz net. The major contributions of this work are the study of active lifetime, the features and behaviors that encourage activity, and the comparison of active lifetime to passive lifetime.

Social informatics is the core of Facebook's business and is its most valuable asset which consists of the social graph and the private data of over 500 million users. However, without secure methods of managing this data, Face book has become vulnerable to privacy risks and devaluation. In Facebook's model, users are asked upon access to grant applications the required permissions without sufficient knowledge of the applications' intentions. As a result, if they are deceived, users risk the exposure of sensitive and personal data. This paper presents a system dubbed FAITH (Face book Applications: Identification, Transformation & Hyper visor) to mitigate or eliminate these issues by enhancing the management of social data. First, FAITH allows users to adjust the visibility of their social informatics for each individual application depending on how much they trust the application. Users can configure FAITH to let non-trusted applications run with the least privileges (least amount of social informatics) to minimize potential privacy leaks. Second, FAITH logs the activities of applications to assist users in making more secure decisions. Users can closely monitor each activity performed by applications to adjust their privacy settings more securely. Third, FAITH allows users to transform their social graph such that different applications see different social graphs preventing the formation of friendship inflation caused by applications. The implementation of FAITH only needs the resources and tools available to the public by Face book and requires no further cooperation from the social network. FAITH is a prototype system: the design and concept can be extended to secure other OSNs (Online Social Networks). Currently, FAITH contains thirteen Face book social applications and has been officially released for public usage with approximately two hundred monthly active users as of now.

Extensive research has been conducted on top of online social networks (OSNs), while little attention has been paid to the data collection process. Due to the large scale of OSNs and their privacy control policies, a partial data set is often used for analysis. The data set analyzed is decided by many factors including the choice of seeds, node selection algorithms, and the sample size. These factors may introduce biases and further contaminate or even skew the results. To evaluate the impact of different factors, this paper examines the OSN graph crawling problem, where the nodes are OSN users and the edges are the links (or relationship) among these users. More specifically, by looking at various factors in the crawling process, the following problems are addressed in this paper: 1) Efficiency: How fast different crawlers discover nodes/links; 2) Sensitivity: How different OSNs and the number of protected users affect crawlers; 3) Bias: How major graph properties are skewed. To the best of our knowledge, our simulations on four real world online social graphs provide the first in-depth empirical answers to these questions.

A user profile on an online social network is characterized by its profile entries (keywords). In this paper, we study the relationship between semantic similarity of user keywords and the social network topology. First, we present a 'forest' model to categorize keywords and define the notion of distance between keywords across multiple categorization trees (i.e., a forest). Second, we use the keyword distance to define similarity functions between a pair of users and show how social network topology can be modeled accordingly. Third, we validate our social network topology model, using a simulated social graph, against a real life social graph dataset.