Unsupervised Graph Representation Learning with Inductive Shallow Node Embedding

Static node embedding algorithms applied to snapshots of real-world applications graphs are unable to capture their evolving process. As a result, the absence of information about the dynamics in these node representations can harm the accuracy and increase processing time of machine learning tasks related to these applications. Aiming at fill the gap regarding the inability of static methods to capture evolving processes on dynamic networks, we propose a biased random walk method named Evolving Node Embedding (EVNE). EVNE leverages the sequential relationship of graph snapshots by incorporating historic information when generating embeddings for the next snapshot. It learns node representations through a neural network, but differs from existing methods as it: (i) incorporates previously run walks at each step; (ii) starts the optimization of the current embedding from the parameters obtained in the previous iteration; and (iii) uses two time-varying parameters to regulate the behavior of the biased random walks over the process of graph exploration. Through a wide set of experiments we show that our approach generates better embeddings, outperforming baselines by up to 20% in a downstream node classification task. EVNE’s embeddings achieve better performance than others, based on experiments with four classifiers and five datasets. In addition, we present seven variations of our model to show the impact of each of EVNE’s mechanisms.

Random walk-based node embedding algorithms learn vector representations of nodes by optimizing an objective function of embedding vectors and skip-bigram statistics computed from random walks on the network. They have demonstrated state-of-the-art performance in many supervised learning problems such as link prediction and node classification. Yet, their properties remain poorly understood. This paper studies properties of random walk based node embeddings in the unsupervised setting of discovering latent network block structure, i.e., learning node representations whose cluster structure in Euclidean space reflects node adjacencies within the network. We characterize the ergodic limits of the embedding objective, its generalization, and related convex relaxations to derive corresponding non-randomized versions of the node embedding objectives. We also characterize the optimal node embedding Grammians of the non-randomized objectives for the expected graph of a two-community Stochastic Block Model (SBM). We prove that the solution Grammian has rank 1 for a suitable nuclear norm relaxation of the non-randomized objective. Comprehensive experimental results on SBM random networks reveal that our non-randomized ergodic objectives yield node embeddings whose distribution is Gaussian-like, centered at the node embeddings of the expected network within each community, and concentrate in the linear degree-scaling regime as the number of nodes increases.

Graph classification is an important problem in many fields, from bioinformatics and neuroscience to computer vision and social network analysis. That said, the task of comparing graphs for the purpose of graph classification faces several major challenges. In particular, an effective graph comparison method must (1) expressively and inductively compare graphs; (2) efficiently compare large graphs; and (3) enable the use of fast machine learning models for graph classification. To address such challenges, we propose Randomized Grid Mapping (RGM), a fast-to-compute feature map that represents a graph via the distribution of its node embeddings in feature space. We justify RGM with close connections to kernel methods: RGM provably approximates the Laplacian kernel mean map and has the multiresolution properties of the pyramid match kernel. We also show that RGM can be extended to incorporate node labels using the Weisfeiler-Lehman framework. Extensive experiments show that graph classification accuracy with RGM feature maps is better than or competitive with many powerful graph kernels, unsupervised graph feature mappings, and deep neural networks. Moreover, comparing graphs based on their node embeddings with RGM is up to an order of magnitude faster than competitive baselines, while maintaining high classification accuracy.

The link prediction problem is a time-evolving model in network science that has simultaneously abetted myriad applications and experienced extensive methodological improvement. Inferring the possibility of emerging links in dynamic social networks, also known as the dynamic link prediction task, is complex and challenging. In contrast to the link prediction in cross-sectional networks, dynamic link prediction methods need to cater to the actor-level temporal changes and associated evolutionary information regarding their micro- (i.e., link formation/deletion) and mesoscale (i.e., community formation) network structure. With the advent of abundant community detection algorithms, the research community has examined community-aware link prediction strategies in static networks. However, the same task in dynamic networks where, apart from the actors and links among them, their community pattern is also dynamic, is yet to be explored. Evolutionary community-aware information, including the associated link structure and temporal neighborhood changes, can effectively be mined to build dynamic similarity metrics for dynamic link prediction. This study aims to develop and integrate such dynamic features with machine learning algorithms for link prediction tasks in dynamic social networks. It also compares the performances of these features against well-known similarity metrics (i.e., ResourceAllocation) for static networks and a time series-based link prediction strategy in dynamic networks. These proposed features achieved high-performance scores, representing them as prospective candidates for both dynamic link prediction tasks and modeling the network growth.

With the ubiquity of complex networks in domains as diverse as social network analysis, recommender systems and epidemiology, improvements in link predictions have a potential for far-reaching impact. In machine learning, a well-known approach to improve classification performance is to combine several methods in an ensemble. It is also well established that, diversity or complementarity of the classifiers to be combined, is a prerequisite for a potential performance improvement of the formed ensemble. In this work, we combine link prediction heuristics (e.g. Common Neighbour, Adamic-Adar etc.), with static graph neural networks, discrete dynamic graph neural networks, and continuous dynamic graph neural networks. We analyze the importance of each of the methods and how complementary they are with respect to each other. We also use greedy searches to select better subsets of methods to combine and train the ensemble in a roll-forward manner to make it adapt to changes in the dynamic network over time. In the conducted experiments on a number of challenging link prediction tasks, the ensemble from the greedy search and the roll-forward ensemble shows on average a 55% and 61% improvement respectively, over the best single method.

BackgroundHighly dynamic social networks, where connectivity continuously changes in time, are becoming more and more pervasive. Knowledge mobilization, which refers to the use of knowledge toward the achievement of goals, is one of the many examples of dynamic social networks. Despite the wide use and extensive study of dynamic networks, their temporal component is often neglected in social network analysis, and statistical measures are usually performed on static network representations. As a result, measures of importance (like betweenness centrality) typically do not reveal the temporal role of the entities involved. Our goal is to contribute to fill this limitation by proposing a form of temporal betweenness measure (foremost betweenness).MethodsOur method is analytical as well as experimental: we design an algorithm to compute foremost betweenness, and we apply it to a case study to analyze a knowledge mobilization network.ResultsWe propose a form of temporal betweenness measure (foremost betweenness) to analyze a knowledge mobilization network and we introduce, for the first time, an algorithm to compute exact foremost betweenness. We then show that this measure, which explicitly takes time into account, allows us to detect centrality roles that were completely hidden in the classical statistical analysis. In particular, we uncover nodes whose static centrality was negligible, but whose temporal role might instead be important to accelerate mobilization flow in the network. We also observe the reverse behavior by detecting nodes with high static centrality, whose role as temporal bridges is instead very low.ConclusionIn this paper, we focus on a form of temporal betweenness designed to detect accelerators in dynamic networks. By revealing potentially important temporal roles, this study is a first step toward a better understanding of the impact of time in social networks and opens the road to further investigation.

This chapter delves into the intersection of social media and public health, specifically focusing on four key methodologies: sentiment analysis, social network analysis, topic modeling, and trend analysis. The scope of the review encompasses these distinct but interconnected methodologies and their applications in the context of public health and the healthcare industry. Our findings indicate that these techniques offer significant potential for extracting valuable insights from social media data, which can inform public health initiatives and policies. Sentiment analysis provides a pulse on public opinion on health-related topics; social network analysis reveals patterns of communication and information dissemination; topic modeling identifies trending health topics on social media platforms; and trend analysis helps in understanding the evolution of these trends over time. The review concludes that the integration of these methodologies with social media data presents a rich avenue for both research and practical applications in public health. As social media continues to evolve, so will the opportunities for public health research and practice, underscoring the need for continued exploration and refinement of these techniques.

Social networks: Ideas for future research from the vantage point of political science

This article examines the application of social network theory in Dutch law enforcement. Increasing amounts of information about habitual lawbreakers and criminal networks are collected under the paradigm of Intelligence-Led Policing. Combined with data gathered from open sources such as social media, such resources allow criminal analysts trained in social network analysis (SNA) at the Police Academy of The Netherlands to apply advanced network analysis methodology and crime scripting. This in turn helps the police to identify crucial weak spots in illicit arrangements and criminal business processes. A case study of the 'Blackbird' crime network, involved in the wholesale cultivation of cannabis is presented to illustrate the power of SNA when combined with crime script analysis. Using a mix of quantitative and qualitative analysis, the topology of the 86-strong Blackbird network is laid out and its substructures and key individuals exposed. In detailing the network's social embeddedness, the authors clarify the importance of female actors for the flexibility and efficiency of the network structure and thereby for the continuity of criminal business. Applying SNA is already helping criminal intelligence units of the Dutch police in identifying intelligence gaps and potential informants. Working in symbiosis, analysts and informant handlers develop a better understanding of strategic targeting and access points to relatively unknown criminal communities and –markets. To be delivered in a timely way to be useful in ongoing criminal investigations, SNA products require even faster data processing. Also, when applied to dark networks SNA should be tailored to better take network dynamics into account, in particular regarding the adaptability to network disruption.

DynHEN: A heterogeneous network model for dynamic bipartite graph representation learning

A social structure constituting a set of nodes, representing social actors and edges or lines representing relation between these nodes or actors is a social network. Social network plays a vital role in circulation of information and innovation leading to analysis of the network and attracted attention in research field. The analysis of social network as a whole means, representation of all its actors and structure present in that social network forming part of a community. Community detection aims to divide the network into dense areas of graph. The dense regions usually correlates to entities which are familiar to each other, and form part of a community. Similar tastes and desires of the members in a community, enables exchange of information amongst various communities. Network science is the study of complex networks. There are various models defined in network science. Erdős–Renyi random graph model, Configuration model, Watts–Strogatz small world model, Barabasi–Albert (BA) preferential attachment model, Mediation-driven attachment model, Fitness model are the various network science models. The current research work consists of application of these models on real world data and comparison between the final output with the use of R Studio.

Over the last 20 years, there has been growing interest in social network analysis across a variety of social science disciplines, particularly in sociology, antropology, communication science, health and medicine, managment and political science. The roots of a social network perspective can be found in Simmel and Moreno discovery that the nature of ties themselves rather than the social group per se that lay at the center of many human behaviors. Despite these early insights, social network analysis gain its prominence only with revived interest in social networks as essence of social structure in 1980s (Burt, Bourdieu, Coleman, Putnam, Castells, et al.). This article aims to explore current metodological social network analysis approaches and how they can be succesfully employed in migration studies. It is argued that previuos dominant quantitative social network analysis based on full network and egocentric network analysis nowadays is renewed by social capital theory and social support approach. This new social network perspective allows to investigate how social networks function, how they can be constructed, maintained and reproduced, how are network dynamics across time and space.

Several activities, comprising animate and inanimate entities, can be examined by means of Social Network Analysis (SNA). Classification tasks within social network structures remain crucial research problems in SNA. Inherent and latent facts about social graphs can be effectively exploited for training Artificial Intelligence (AI) models in a bid to categorize actors/nodes as well as identify clusters with respect to a given social network. Thus, important factors such as the individual attributes of spatial social actors and the underlying patterns of relationship binding these social actors must be taken into consideration. These factors are relevant to understanding the nature and dynamics of a given social graph. In this paper, we have proposed a hybrid model: Representation Learning via Knowledge-Graph Embeddings and Convolution Operations (RLVECO) which has been modelled for studying and extracting meaningful facts from social network structures to aid in node classification and community detection problems. RLVECO utilizes an edge sampling approach for exploiting features of a social graph, via learning the context of each actor with respect to its neighboring actors, with the aim of generating vector-space embeddings per actor which are further exploited for unexpressed representations via a sequence of convolution operations. Successively, these relatively low-dimensional representations are fed as input features to a downstream classifier for solving community detection and node classification problems about a given social network.

Social media networks are dynamic. As such, the order in which network ties develop is an important aspect of the network dynamics.This study proposes a novel dynamic network model, the Nodal Attribute-based Temporal Exponential Random Graph Model (NATERGM) for dynamic network analysis. The proposed model focuses on how the nodal attributes of a network affect the order in which the network ties develop. Empirical results showed that the NATERGM demonstrated an enhanced pattern testing capability compared to benchmark models. The proposed NATERGM model helps explain the roles of nodal attributes in the formation process of dynamic networks.

BackgroundRapid identification of subject experts for medical topics helps in improving the implementation of discoveries by speeding the time to market drugs and aiding in clinical trial recruitment, etc. Identifying such people who influence opinion through social network analysis is gaining prominence. In this work, we explore how to combine named entity recognition from unstructured news articles with social network analysis to discover opinion leaders for a given medical topic.MethodsWe employed a Conditional Random Field algorithm to extract three categories of entities from health-related new articles: Person, Organization and Location. We used the latter two to disambiguate polysemy and synonymy for the person names, used simple rules to identify the subject experts, and then applied social network analysis techniques to discover the opinion leaders among them based on their media presence. A network was created by linking each pair of subject experts who are mentioned together in an article. The social network analysis metrics (including centrality metrics such as Betweenness, Closeness, Degree and Eigenvector) are used for ranking the subject experts based on their power in information flow.ResultsWe extracted 734,204 person mentions from 147,528 news articles related to obesity from January 1, 2007 through July 22, 2010. Of these, 147,879 mentions have been marked as subject experts. The F-score of extracting person names is 88.5%. More than 80% of the subject experts who rank among top 20 in at least one of the metrics could be considered as opinion leaders in obesity.ConclusionThe analysis of the network of subject experts with media presence revealed that an opinion leader might have fewer mentions in the news articles, but a high network centrality measure and vice-versa. Betweenness, Closeness and Degree centrality measures were shown to supplement frequency counts in the task of finding subject experts. Further, opinion leaders missed in scientific publication network analysis could be retrieved from news articles.