Social Network Analysis in Streaming Call Graphs

Knowledge graphs have emerged as a key abstraction for organizing information in diverse domains and their embeddings are increasingly used to harness their information in various information retrieval and machine learning tasks. However, the ever growing size of knowledge graphs requires computationally efficient algorithms capable of scaling to graphs with millions of nodes and billions of edges. This paper presents DGL-KE, an open-source package to efficiently compute knowledge graph embeddings. DGL-KE introduces various novel optimizations that accelerate training on knowledge graphs with millions of nodes and billions of edges using multi-processing, multi-GPU, and distributed parallelism. These optimizations are designed to increase data locality, reduce communication overhead, overlap computations with memory accesses, and achieve high operation efficiency. Experiments on knowledge graphs consisting of over 86M nodes and 338M edges show that DGL-KE can compute embeddings in 100 minutes on an EC2 instance with 8 GPUs and 30 minutes on an EC2 cluster with 4 machines with 48 cores/machine. These results represent a 2× ~ 5× speedup over the best competing approaches. DGL-KE is available on https://github.com/awslabs/dgl-ke.

With the availability of multicore and manycore systems with large main memories, Symmetric Multiprocessors (SMPs) can process large scale problems such as graphs spanning millions of nodes and billions of edges which used to be the domain of large clusters. Due to the shared memory architecture of SMPs, we can apply fast, novel policies such as Transactional Memory (TM) to speed up the processing of such problems. Interestingly, most applications in bioinformatics, social networks, and cybersecurity can be represented as large graphs. However, these graphs are sparse in nature. Therefore, TM as a synchronization policy for critical sections can provide better performance since it performs better in low conflict scenarios. Many TM variants exist including: Software, Hardware and combinations of both. Furthermore, a TM can adapt to applications' behavior. In this paper, we introduce DAdHTM. A Dynamically Adaptive Hardware TM (DAdHTM) designed to adapt the HTM to the application's behavior. DAdHTM outperforms not only locks but also low overhead Software TM (STM) and even Adaptive Hybrid TM (DyAdHyTM) on a 48 core 64GB memory dual socket SMP machine. Our experiments show that DAdHTM has low overhead. It is competitive at core/thread counts below 24 (close to the best) and outperforms all other techniques at core/thread counts above 24 by up to 43%.

Event detection in wireless sensor networks is a sophisticated method for processing sampled data directly on the sensor nodes, thereby reducing the need for multihop communication with the base station of the network. In contrast to application-agnostic compression or aggregation techniques, event detection pushes application-level knowledge into the network. In-network event detection - especially the distributed form involving multiple sensor nodes - has thus an exceptional potential to increase energy efficiency, thus prolonging the lifetime of the network. In this article, we summarize recently proposed system architectures and algorithms employed for event detection in wireless sensor networks. On the example of the AVS-Extrem platform, we illustrate how energy-efficient event detection can be implemented through a combination of custom hardware design and distributed event detection algorithms. We then continue to present a brief evaluation of the detection accuracy and the energy consumption that is achievable by current systems.

We consider stochastic transition matrices from large social and information networks. For these matrices, we describe and evaluate three fast methods to estimate one column of the matrix exponential. The methods are designed to exploit the properties inherent in social networks, such as a power-law degree distribution. Using only this property, we prove that one of our three algorithms has a sublinear runtime. We present further experimental evidence showing that all three of them run quickly on social networks with billions of edges, and they accurately identify the largest elements of the column.

With the advent of social networks and the rise of mobile technologies, users have become ubiquitous sensors capable of monitoring various real-world events in a crowd-sourced manner. Location-based social networks have proven to be faster than traditional media channels in reporting and geo-locating breaking news, i.e. Osama Bin Laden’s death was first confirmed on Twitter even before the announcement from the communication department at the White House. However, the deluge of user-generated data on these networks requires intelligent systems capable of identifying and characterizing such events in a comprehensive manner. The data mining community coined the term, event detection , to refer to the task of uncovering emerging patterns in data streams . Nonetheless, most data mining techniques do not reproduce the underlying data generation process, hampering to self-adapt in fast-changing scenarios. Because of this, we propose a probabilistic machine learning approach to event detection which explicitly models the data generation process and enables reasoning about the discovered events. With the aim to set forth the differences between both approaches, we present two techniques for the problem of event detection in Twitter : a data mining technique called Tweet-SCAN and a machine learning technique called Warble. We assess and compare both techniques in a dataset of tweets geo-located in the city of Barcelona during its annual festivities. Last but not least, we present the algorithmic changes and data processing frameworks to scale up the proposed techniques to big data workloads.

In this paper, we present a multi-level empirical study of locality structures of several temporal call graphs containing both mobile and fixed-line call graphs emphasizing on comparing the patterns of these two types of call graphs. We investigate the topological patterns of the cohesive subgraphs in a mesoscopic scale, and we find several novel patterns in these call graphs. We study the correlation between the link weights and their localities. To our surprise, we can find there is nearly no correlation between link weights and their edge betweenness values. We also find that in the network of communities, communities’ sizes and betweenness centralities are highly positively correlated, which indicates large communities tend to be in the center of the call networks. Our analysis also suggests that ‘small-world’ phenomenon still exists in the community-based networks. We believe that our analysis results will help Telecom operators have better understanding of their customers.

Event detection techniques are crucial for environmental monitoring and object tracking applications in wireless sensor networks. Event detection requires sensor readings to be collected from multiple sensors, and as sensors have limited resources, their readings should be retrieved efficiently. Existing aggregation-based event detection methods, however, require all sensors in the network to transmit their readings, which results in high energy consumption for the network. We propose an energy-efficient event detection technique for estimating the state of the whole environment based on only some of the sensor readings and use a contour map to represent the outline of the environment. To detect events in 3D environments, we modify an existing 2D contour mapping algorithm, extending it for 3D environments. By simulation, we show how our event detection technique is more energy-efficient than existing solutions that take readings from all sensors. We evaluated our method from the point of energy efficiency and found that it improves the energy efficiency of event detection in 3D wireless sensor networks.

partitioning graphs at extent is a critical contest for several solicitations that incorporate provision a graph transversely disk, technologies, or data centers. Graph partitioning is an exact fine, considered difficult with satirical collected works, but surviving algorithms classically cannot measure to billions of edges, or cannot deliver assurances about partition bulks. In this work, we announce an effective algorithm, composed label propagation, for exactly partitioning immense graphs while hungrily exploiting edge vicinity, the expanse of edges that are allocated to the equal sliver of a partition. By merging the computational competence of label propagation where nodes are iteratively relabeled to the similar ‘label’ as the variety of their graph neighbors' - with the assurances of inhibited optimization -managing the propagation by an undeviating platform obliging the partition scopes - our algorithm types it virtually imaginable to divide graphs with billions of edges. Our algorithm is attracted by the competition of finishing Graph assessments in a detached structure. Because this requires conveying each node in a graph to a physical machine with memory boundaries, it is disapprovingly essential to guarantee the resultant partition debris do not exceed several single machines. We estimate our algorithm for its dividing performance on the social graph, and also learning its enactment when dividing Facebook's ‘People You May Know’ service (PYMK), the dispersed scheme answerable for the article abstraction and the standing of the friends-of-friends of all vigorous Facebook users. In a live placement, we observed typical interrogation times and average network circulation stages that were 50.5% and 37.1% (respectively) when matched to the earlier naive random sharding.

Convergence is a term that has been gaining increasing attention across the globe. Conceptually situated between technological and socio-cultural domains – as has been intensely debated between the technological and social determinists since the birth of computer mediated communication (CMC) – mobile culture is no exception. It is powered by inter-operable and multi-functional mobile telecommunication technologies, which are also becoming increasingly high functional especially via fixed-mobile convergence, interconnecting fixed and wireless broadband networks. Wide diffusion and convergence of digital communication technologies – wired and wireless – in recent years have resulted in the emergence of two key paradigmatic convergences: between technology and sociality, and between productivity and play. I will first elucidate these convergences; I will then discuss cultural characteristics of specific East Asian cultures, focusing on South Korea, China, and Japan.

Understanding the Design of Mobile Social Networking

Event detection in social networks

IMPC: Influence maximization based on multi-neighbor potential in community networks

VCE as commercial monitoring service provider has gained considerable experience in the topic from more than 600 projects worldwide. Data management, processing, analysis and interpretation are always a challenge especially in case of permanent monitoring. Monitoring systems with hundreds of sensors produce gigabytes of measurement data every day. Full automatic operation and analysis is required to provide the results in time especially in case of alert systems. The challenges for these systems are: - Elimination of noise - Reliable detection and removal of monitoring errors - Data management (transfer, storage, archiving) - Data pre-analysis - Detection of events - Correlation analysis and cross checking - Statistical analysis - Data analysis and diagnosis - Triggering of action (alerts, automatic traffic management) - Access to data - Data presentation - Support of systems operators and engineers for decision making KEYWORDS Web-User-Interface, automatic data treatment, alert systems, data management, event detection, statistical analysis.

Advertising in social network has become a multi-billion-dollar industry. A main challenge is to identify key influencers who can effectively contribute to the dissemination of information. Although the influence maximization problem, which finds a seed set of k most influential users based on certain propagation models, has been well studied, it is not target-aware and cannot be directly applied to online advertising. In this paper, we propose a new problem, named Keyword-Based Targeted Influence Maximization (KB-TIM), to find a seed set that maximizes the expected influence over users who are relevant to a given advertisement. To solve the problem, we propose a sampling technique based on weighted reverse influence set and achieve an approximation ratio of (1 − 1/ e − ε ). To meet the instant-speed requirement, we propose two disk-based solutions that improve the query processing time by two orders of magnitude over the state-of-the-art solutions, while keeping the theoretical bound. Experiments conducted on two real social networks confirm our theoretical findings as well as the efficiency. Given an advertisement with 5 keywords, it takes only 2 seconds to find the most influential users in a social network with billions of edges.

Online monitoring of multivariate water quality data is becoming a practical means of improving distribution network management and meeting water security goals. Changes in water quality are often due to changes in the hydraulic operations of the network. These operational changes create patterns of water quality change that are similar, but not exactly the same, from one instance to the next. Classification of multivariate change patterns through trajectory clustering is introduced in this paper to create a pattern library from historical water quality data and as an online process with the goal of reducing false positive water quality event detections. Prior to event declaration, a short sequence of the preceding multivariate data is compared against the pattern library to assess its similarity to a previously observed pattern. A fuzzy clustering algorithm is utilized to assign multivariate pattern memberships for water quality patterns associated with water quality events in both the offline an...