Faraz Zaidi

Maritime constellations: a complex network approach to shipping and ports

The analysis of community structures is one major research field in the science of networks. This exercise is often biased by strong hierarchical configurations as it is the case in container shipping. After reviewing the multiple definitions of port systems, this paper applies a topological decomposition method to worldwide inter-port maritime links. Isolating ports of comparable size reveals hidden substructures with the help of graph visualization. While geographic proximity is one main explanatory factor in the emergence of port systems, other logics also appear, such as specialized and long-distance trading links. This research provides interesting evidence about the role of geography, technology and trade in the architecture of maritime networks.

Model for generating artificial social networks having community structures with small-world and scale-free properties

Recent interest in complex systems and specially social networks has catalyzed the development of numerous models to help understand these networks. A number of models have been proposed recently where they are either variants of the small-world model, the preferential attachment model or both. Three fundamental properties attributed to identify these complex networks are high clustering coefficient, small average path length and the vertex connectivity following power-law distribution. Different models have been presented to generate networks having all these properties. In this study, we focus on social networks and another important characteristic of these networks, which is the presence of community structures. Often misinterpret with the metric called clustering coefficient, we first show that the presence of community structures is indeed different from having high clustering coefficient. We then define a new network generation model which exhibits all the fundamental properties of complex networks along with the presence of community structures.

Are All Social Networks Structurally Similar

The modern age has seen an exponential growth of social network data available on the web. Analysis of these networks reveal important structural information about these networks in particular and about our societies in general. More often than not, analysis of these networks is concerned in identifying similarities among social networks and how they are different from other networks such as protein interaction networks, computer networks and food web. In this paper, our objective is to perform a critical analysis of different social networks using structural metrics in an effort to highlight their similarities and differences. We use five different social network datasets which are contextually and semantically different from each other. We then analyze these networks using a number of different network statistics and metrics. Our results show that although these social networks have been constructed from different contexts, they are structurally similar.

Tunable and Growing Network Generation Model with Community Structures

Recent years have seen a growing interest in the modeling and simulation of social networks to understand several social phenomena. Two important classes of networks, small world and scale free networks have gained a lot of research interest. Another important characteristic of social networks is the presence of community structures. Many social processes such as information diffusion and disease epidemics depend on the presence of community structures making it an important property for network generation models to be incorporated. In this paper, we present a tunable and growing network generation model with small world and scale free properties as well as the presence of community structures. The major contribution of this model is that the communities thus created satisfy three important structural properties: connectivity within each community follows power-law, communities have high clustering coefficient and hierarchical community structures are present in the networks generated using the proposed model. Furthermore, the model is highly robust and capable of producing networks with a number of different topological characteristics varying clustering coefficient and inter-cluster edges. Our simulation results show that the model produces small world and scale free networks along with the presence of communities depicting real world societies and social networks.

Empirical Analysis of Seed Selection Criterion in Influence Mining for Different Classes of Networks

Recent years have seen social networks gain lot of popularity to share information, connecting millions of people from all over the world. Studying the spread of information, or Information Diffusion in these networks has shaped into a well known field of study with numerous applications in areas such as marketing, politics, and personality evaluation. Researchers have studied information diffusion under various models and opted centrality-based algorithms that offer better results over many other approaches. These algorithms try to select initial seed nodes effectively so as to maximize influence in a network in minimum time. However, since different networks follow different structural properties, motivating the need to study different diffusion strategies for networks with different structural properties. In this paper, we aim to empirically analyze four different measures of centrality to select seed vertices for influence mining on four classes of networks: small-World networks, scale-free networks, small world-scale free networks and random networks. These networks are generated equivalent in size to four semantically different real world social networks. We use two most frequently used diffusion models: Independent Cascade model and Linear Threshold model for analysis. Our results show interesting behavior of various centrality measures for the above said classes of networks.

Demographic and Structural Characteristics to Rationalize Link Formation in Online Social Networks

Recent years have seen tremendous growth of many online social networks such as Facebook, Linked In and My Space. People connect to each other through these networks forming large social communities providing researchers rich datasets to understand, model and predict social interactions and behaviors. New contacts in these networks can be formed either due to an individuals demographic profile such as age group, gender, geographic location or due to networks structural dynamics such as triadic closure and preferential attachment, or a combination of both demographic and structural characteristics. A number of network generation models have been proposed in the last decade to explain the structure, evolution and processes taking place in different types of networks, and notably social networks. Network generation models studied in the literature primarily consider structural properties, and in some cases an individuals demographic profile in the formation of new social contacts. These models do not present a mechanism to combine both structural and demographic characteristics for the formation of new links. In this paper, we propose a new network generation algorithm which incorporates both these characteristics to model growth of a network. We use different publicly available Facebook datasets as benchmarks to demonstrate the correctness of the proposed network generation model.

Topology of Complex Networks and Performance Limitations of Community Detection Algorithms

One of the most widely studied problems in the analysis of complex networks is the detection of community structures. Many algorithms have been proposed to find communities but the quest to find the best algorithm is still on. More often than not, researchers focus on developing fast and accurate algorithms that can be generically applied to networks from various domains. As the topology of networks changes with respect to domains, community detection algorithms fail to accommodate these changes to detect communities. In this paper, we attempt to highlight this problem by studying networks with different topologies and evaluate the performance of community detection algorithms in the light of these topological changes. To generate networks with different topologies, we used the well-known Lancichinetti–Fortunato–Radicchi (LFR) model, and we also propose a new model named Naïve Scale-Free Clustering to avoid any bias that can be introduced by the underlying network generation model. Results reveal several limitations of the current popular network clustering algorithms failing to correctly find communities. This suggests the need to revisit the design of current clustering algorithms in order to improve their performances.

A Graph-Based Method for Detecting Rare Events: Identifying Pathologic Cells

Detection of outliers and anomalous behavior is a well-known problem in the data mining and statistics fields. Although the problem of identifying single outliers has been extensively studied in the literature, little effort has been devoted to detecting small groups of outliers that are similar to each other but markedly different from the entire population. Many real-world scenarios have small groups of outliers--for example, a group of students who excel in a classroom or a group of spammers in an online social network. In this article, the authors propose a novel method to solve this challenging problem that lies at the frontiers of outlier detection and clustering of similar groups. The method transforms a multidimensional dataset into a graph, applies a network metric to detect clusters, and renders a representation for visual assessment to find rare events. The authors tested the proposed method to detect pathologic cells in the biomedical science domain. The results are promising and confirm the available ground truth provided by the domain experts.

Resilience of Social Networks under Different Attack Strategies

Recent years have seen the world become a closely connected society with the emergence of different types of social networks. Online social networks have provided a way to bridge long distances and establish numerous communication channels which were not possible earlier. These networks exhibit interesting behavior under intentional attacks and random failures where different structural properties influence the resilience in different ways. n nIn this paper, we perform two sets of experiments and draw conclusions from the results pertaining to the resilience of social networks. The first experiment performs a comparative analysis of four different classes of networks namely small world networks, scale free networks, small world-scale free networks and random networks with four semantically different social networks under different attack strategies. The second experiment compares the resilience of these semantically different social networks under different attack strategies. Empirical analysis reveals interesting behavior of different classes of networks with different attack strategies.