Online Social Network Analysis Research Articles

Exploring social networks through stochastic multilayer graph modeling

Several graph models are available today to model online social networks. These graph models are used to analyze the structural properties of the online social network, such as detecting communities, finding the influential spreader and predicting the behavior of the network. However, these models are based on deterministic single-layer graphs that may not be appropriate when online users use multiple social networks at the same time and social networks provide specific services. Moreover, because of the unknown and dynamic nature related to the behaviors and activities of online users, as well as structural and behavioral parameters, which may vary over time, stochastic multi-layer models could be applied to better capture and represent this phenomenon, as well as the dynamic nature of social networks. For example, in recommender systems, users' interests are unknown parameters and vary over time. Therefore, stochastic multilayer graph modeling can be used to develop recommender systems by considering different layers for different types of interests or preferences. In this paper, we propose a stochastic multilayer graph in which the edges are associated with random variables as a potential graph model for the analysis of online social networks. Therefore, after redefine some network measures related to stochastic multilayer graphs, we propose a Cellular Goore Game (CGG) based algorithm to computes the redefine network measures. A CGG-based algorithm computes defined network measures by learning automata from the edges of stochastic multilayer graphs. The experimental results show that the new CGG-based algorithm requires fewer samples from the edges of stochastic multilayer graphs than the standard sampling method in network measures calculation. Furthermore, the obtained results demonstrate that, from an evaluation perspective, the CGG-based algorithm provides superior results in terms of Kolmogorov-Smirnov (KS-test), Pearson Correlation Coefficient (PCC), Normalized Root Mean Square Error (NRMSE) and Kullback–Leibler divergence (KL-test).

Suicidal Ideation Detection from social media: A Detailed Review of Machine Learning and Deep Learning

Social networks are crucial tools for learning about people's attitudes towards various issues since they allow them to express their thoughts to friends and family. Recent years have seen considerable challenges in natural language processing (NLP) and psychology regarding the detection of suicidal ideation via online social network analysis. The complex early signs of suicide ideations can be identified with the proper use of social media information, which can afterwards save many lives. Even though numerous strategies have been used over the past few decades to identify suicidal thoughts, machine learning (ML) and deep learning (DL) methods offer more insightful results. So, in this study, we examine several cutting-edge ML and DL approaches for detecting suicidal ideation. We also observe a few significant issues in the entire corpus of literature that could be explored in further research. Finally, we hope this study will shed light on critical issues in identifying suicidal thoughts on social media for readers and ML and DL researchers.

Guest Editorial: Blockchain and decentralised solutions for social good

Guest Editorial: Blockchain and decentralised solutions for social good

Characterizing the Discourse of Popular Diets to Describe Information Dispersal and Identify Leading Voices, Interaction, and Themes of Mental Health: Social Network Analysis

Social media has transformed the way health messages are communicated. This has created new challenges and ethical considerations while providing a platform to share nutrition information for communities to connect and for information to spread. However, research exploring the web-based diet communities of popular diets is limited. This study aims to characterize the web-based discourse of popular diets, describe information dissemination, identify influential voices, and explore interactions between community networks and themes of mental health. This exploratory study used Twitter social media posts for an online social network analysis. Popular diet keywords were systematically developed, and data were collected and analyzed using the NodeXL metrics tool (Social Media Research Foundation) to determine the key network metrics (vertices, edges, cluster algorithms, graph visualization, centrality measures, text analysis, and time-series analytics). The vegan and ketogenic diets had the largest networks, whereas the zone diet had the smallest network. In total, 31.2% (54/173) of the top users endorsed the corresponding diet, and 11% (19/173) claimed a health or science education, which included 1.2% (2/173) of dietitians. Complete fragmentation and hub and spoke messaging were the dominant network structures. In total, 69% (11/16) of the networks interacted, where the ketogenic diet was mentioned most, with depression and anxiety and eating disorder words most prominent in the "zone diet" network and the least prominent in the "soy-free," "vegan," "dairy-free," and "gluten-free" diet networks. Social media activity reflects diet trends and provides a platform for nutrition information to spread through resharing. A longitudinal exploration of popular diet networks is needed to further understand the impact social media can have on dietary choices. Social media training is vital, and nutrition professionals must work together as a community to actively reshare evidence-based posts on the web.

Identification of early symptoms of endometriosis through the analysis of online social networks: A social media study.

Endometriosis is a complex full-body inflammation disease with an average time to diagnosis of 7-10 years. Social networks give opportunity to patient to openly discuss about their condition, share experiences, and seek advice. Thus, data from social media may provide insightful data about patient's experience. This study aimed at applying a text-mining approach to online social networks in order to identify early signs associated with endometriosis. An automated exploration technique of online forums was performed to extract posts. After a cleaning step of the built corpus, we retrieved all symptoms evoked by women, and connected them to the MedDRA dictionary. Then, temporal markers allowed targeting only the earliest symptoms. The latter were those evoked near a marker of precocity. A co-occurrence approach was further applied to better account for the context of evocations. Results were visualised using the graph-oriented database Neo4j. We collected 7148 discussions threads and 78,905 posts from 10 French forums. We extracted 41 groups of contextualised symptoms, including 20 groups of early symptoms associated with endometriosis. Among these groups of early symptoms, 13 were found to portray already known signs of endometriosis. The remaining 7 clusters of early symptoms were limb oedema, muscle pain, neuralgia, haematuria, vaginal itching, altered general condition (i.e. dizziness, fatigue, nausea) and hot flush. We pointed out some additional symptoms of endometriosis qualified as early symptoms, which can serve as a screening tool for prevention and/or treatment purpose. The present findings offer an opportunity for further exploration of early biological processes triggering this disease.

Online social network user performance prediction by graph neural networks

Online social networks provide rich information that characterizes the user’s personality, his interests, hobbies, and reflects his current state. Users of social networks publish photos, posts, videos, audio, etc. every day. Online social networks (OSN) open up a wide range of research opportunities for scientists. Much research conducted in recent years using graph neural networks (GNN) has shown their advantages over conventional deep learning. In particular, the use of graph neural networks for online social network analysis seems to be the most suitable. In this article we studied the use of graph convolutional neural networks with different convolution layers (GCNConv, SAGEConv, GraphConv, GATConv, TransformerConv, GINConv) for predicting the user’s professional success in VKontakte online social network, based on data obtained from his profiles. We have used various parameters obtained from users’ personal pages in VKontakte social network (the number of friends, subscribers, interesting pages, etc.) as their features for determining the professional success, as well as networks (graphs) reflecting connections between users (followers/ friends). In this work we performed graph classification by using graph convolutional neural networks (with different types of convolution layers). The best accuracy of the graph convolutional neural network (0.88) was achieved by using the graph isomorphism network (GIN) layer. The results, obtained in this work, will serve for further studies of social success, based on metrics of personal profiles of OSN users and social graphs using neural network methods.

GTIP: A Gaming-Based Topic Influence Percolation Model for Semantic Overlapping Community Detection.

Community detection in semantic social networks is a crucial issue in online social network analysis, and has received extensive attention from researchers in various fields. Different conventional methods discover semantic communities based merely on users’ preferences towards global topics, ignoring the influence of topics themselves and the impact of topic propagation in community detection. To better cope with such situations, we propose a Gaming-based Topic Influence Percolation model (GTIP) for semantic overlapping community detection. In our approach, community formation is modeled as a seed expansion process. The seeds are individuals holding high influence topics and the expansion is modeled as a modified percolation process. We use the concept of payoff in game theory to decide whether to allow neighbors to accept the passed topics, which is more in line with the real social environment. We compare GTIP with four traditional (GN, FN, LFM, COPRA) and seven representative (CUT, TURCM, LCTA, ACQ, DEEP, BTLSC, SCE) semantic community detection methods. The results show that our method is closer to ground truth in synthetic networks and has a higher semantic modularity in real networks.

A Comparative Analysis on Suicidal Ideation Detection Using NLP, Machine, and Deep Learning

Social networks are essential resources to obtain information about people’s opinions and feelings towards various issues as they share their views with their friends and family. Suicidal ideation detection via online social network analysis has emerged as an essential research topic with significant difficulties in the fields of NLP and psychology in recent years. With the proper exploitation of the information in social media, the complicated early symptoms of suicidal ideations can be discovered and hence, it can save many lives. This study offers a comparative analysis of multiple machine learning and deep learning models to identify suicidal thoughts from the social media platform Twitter. The principal purpose of our research is to achieve better model performance than prior research works to recognize early indications with high accuracy and avoid suicide attempts. We applied text pre-processing and feature extraction approaches such as CountVectorizer and word embedding, and trained several machine learning and deep learning models for such a goal. Experiments were conducted on a dataset of 49,178 instances retrieved from live tweets by 18 suicidal and non-suicidal keywords using Python Tweepy API. Our experimental findings reveal that the RF model can achieve the highest classification score among machine learning algorithms, with an accuracy of 93% and an F1 score of 0.92. However, training the deep learning classifiers with word embedding increases the performance of ML models, where the BiLSTM model reaches an accuracy of 93.6% and a 0.93 F1 score.

A Novel Multilayer Model for Link Prediction in Online Social Networks Based on Reliable Paths

Link Prediction (LP) is one of the critical problems in Online Social Networks (OSNs) analysis. LP is a technique for predicting forthcoming or missing links based on current information in the OSN. Typically, modelling an OSN platform is done in a single-layer scheme. However, this is a limitation which might lead to incorrect descriptions of some real-world details. To overcome this limitation, this paper presents a multilayer model of OSN for the LP problem by analysing Twitter and Foursquare networks. LP in multilayer networks involves performing LP on a target layer benefitting from the structural information of the other layers. Here, a novel criterion is proposed, which calculates the similarity between users by forming intralayer and interlayer links in a two-layer network (i.e. Twitter and Foursquare). Particularly, LP in the Foursquare layer is done by considering the two-layer structural information. In this paper, according to the available information from the Twitter and Foursquare OSNs, a weighted graph is created and then various topological features are extracted from it. Based on the extracted features, a database with two classes of link existence and no link has been created, and therefore the problem of LP has become a two-class classification problem that can be solved by supervised learning methods. To prove the better performance of the proposed method, Katz and FriendLink indices as well as SEM-Path algorithm have been used for comparison. Evaluations results show that the proposed method can predict new links with better precision.

Influence Maximization Based on Backward Reasoning in Online Social Networks

Along with the rapid development of information technology, online social networks have become more and more popular, which has greatly changed the way of information diffusion. Influence maximization is one of the hot research issues in online social network analysis. It refers to mining the most influential top-K nodes from an online social network to maximize the final propagation of influence in the network. The existing studies have shown that the greedy algorithms can obtain a highly accurate result, but its calculation is time-consuming. Although heuristic algorithms can improve efficiency, it is at the expense of accuracy. To balance the contradiction between calculation accuracy and efficiency, we propose a new framework based on backward reasoning called Influence Maximization Based on Backward Reasoning. This new framework uses the maximum influence area in the network to reversely infer the most likely seed nodes, which is based on maximum likelihood estimation. The scheme we adopted demonstrates four strengths. First, it achieves a balance between the accuracy of the result and efficiency. Second, it defines the influence cardinality of the node based on the information diffusion process and the network topology structure, which guarantees the accuracy of the algorithm. Third, the calculation method based on message-passing greatly reduces the computational complexity. More importantly, we applied the proposed framework to different types of real online social network datasets and conducted a series of experiments with different specifications and settings to verify the advantages of the algorithm. The results of the experiments are very promising.

Tutorials

The tutorials presented at ICWSM 2013 included Advanced Methods for Collecting Social Science Data in the Social Media Field, presented by Riki Conrey; Information-Theoretic Tools for Social Media Analysis, presented by Greg Ver Steeg and Aram Galstyan; Crisis Mapping, Citizen Sensing, and Social Media Analytics: Leveraging Citizen Roles for Crisis Response, presented by Amit Sheth, Patrick Meier, Carlos Castillo, and Hemant Purohit; Multiple Network Models for Complex Online Social Network Analysis, presented by Matteo Magnani and Luca Rossi; and Pulse of Virtual Worlds: Behavioral Mining and SNA in Massive Online Games, presented by Muhammad Ahmad and Jaideep Srivastava.

Network Subgraph Pattern Mining

Mining local connection patterns in large-scale networks is of great significance for understanding the structure and function of biological networks and online social networks. In practical applications, there are many problems and challenges, such as unknown network structure, large network scale, large number of subgraphs, and large operation of subgraph pattern analysis, which make it very difficult to analyze the local connection patterns of large-scale network graph accurately and quickly. In order to solve this problem, this paper designs a method of crawler and sampling to obtain the topology of the unknown network. At the same time, the deviation introduced in the process of data acquisition is modeled and analyzed, which can be compensated and corrected. A sampling unbiased estimation method suitable for large-scale static graph and high-speed dynamic flow graph is designed. The research results of this paper can be used to accurately and quickly mine and estimate the eigenvalues of local connection patterns, and provide important technical means for computer network traffic monitoring, online social network analysis and biomolecular network information mining.

Uncovering information diffusion patterns in different networks using the L-metric

ABSTRACT Information diffusion is an important branch of online social network analysis. In this paper, we construct a new metric, the proportion of leaf nodes in a diffusion tree (L-metric), to quantify information diffusion patterns, and we study the impact of the network category and information content on these patterns. Simulation-based experimental studies of real-world social networks show that information diffusion exhibits different patterns in different networks, and niche information does not typically propagate easily in any type of network. These conclusions provide a new perspective for further research on management decisions with regard to online social networks.

A novel approach for efficient stance detection in online social networks with metaheuristic optimization

In the 19th and 20th centuries, social networks have been an important topic in a wide range of fields from sociology to education. However, with the advances in computer technology in the 21st century, significant changes have been observed in social networks, and conventional networks have evolved into online social networks. The size of these networks, along with the large amount of data they generate, has introduced new social networking problems and solutions. Social network analysis methods are used to understand social network data. Today, several methods are implemented to solve various social network analysis problems, albeit with limited success in certain problems. Thus, the researchers develop new methods or recommend solutions to improve the performance of the existing methods. In the present paper, a novel optimization method that aimed to classify social network analysis problems was proposed. The problem of stance detection, an online social network analysis problem, was first tackled as an optimization problem. Furthermore, a new hybrid metaheuristic optimization algorithm was proposed for the first time in the current study, and the algorithm was compared with various methods. The analysis of the findings obtained with accuracy, precision, recall, and F-measure classification metrics demonstrated that our method performed better than other methods.

Geographical Structural Features of the WeChat Social Networks

Recently, spatial interaction analysis of online social networks has become a big concern. Early studies of geographical characteristics analysis and community detection in online social networks have shown that nodes within the same community might gather together geographically. However, the method of community detection is based on the idea that there are more links within the community than that connect nodes in different communities, and there is no analysis to explain the phenomenon. The statistical models for network analysis usually investigate the characteristics of a network based on the probability theory. This paper analyzes a series of statistical models and selects the MDND model to classify links and nodes in social networks. The model can achieve the same performance as the community detection algorithm when analyzing the structure in the online social network. The construction assumption of the model explains the reasons for the geographically aggregating of nodes in the same community to a degree. The research provides new ideas and methods for nodes classification and geographic characteristics analysis of online social networks and mobile communication networks and makes up for the shortcomings of community detection methods that do not explain the principle of network generation. A natural progression of this work is to geographically analyze the characteristics of social networks and provide assistance for advertising delivery and Internet management.

The Online Social Networks Analysis: State of the Art

The Online Social Networks Analysis: State of the Art

A social network based approach to identify and rank influential nodes for smart city

Purpose In online social network analysis, the problem of identification and ranking of influential nodes based on their prominence has attracted immense attention from researchers and practitioners. Identification and ranking of influential nodes is a challenging problem using Twitter, as data contains heterogeneous features such as tweets, likes, mentions and retweets. The purpose of this paper is to perform correlation between various features, evaluation metrics, approaches and results to validate selection of features as well as results. In addition, the paper uses well-known techniques to find topical authority and sentiments of influential nodes that help smart city governance and to make importance decisions while understanding the various perceptions of relevant influential nodes. Design/methodology/approach The tweets fetched using Twitter API are stored in Neo4j to generate graph-based relationships between various features of Twitter data such as followers, mentions and retweets. In this paper, consensus approach based on Twitter data using heterogeneous features has been proposed based on various features such as like, mentions and retweets to generate individual list of top-k influential nodes based on each features. Findings The heterogeneous features are meant for integrating to accomplish identification and ranking tasks with low computational complexity, i.e. O(n), which is suitable for large-scale online social network with better accuracy than baselines. Originality/value Identified influential nodes can act as source in making public decisions and their opinion give insights to urban governance bodies such as municipal corporation as well as similar organization responsible for smart urban governance and smart city development.

A Machine Learning-based Online Social Network Analysis for 360-degree User Profiling

This paper aims to analyse the online social network for reconnaissance of people for finding their potentiality. The work considers one of the famous social networking sites, Twitter, where people express their thoughts and ideas, through which the people in the site knowingly or unknowingly reveal the information about themselves such as personal interests, likes and dislikes. The Machine Learning technique facilitates the work to mine the tweet data of a person to get his/her 360-degree profiling. This profiling is helpful to identify the personality type of a person, which is essential for the Government to identify the people involved in spreading the fake news in Twitter.

A Study Regarding Human Perception of a Company through Social Media

Of the utmost importance is finding the social networks that best fit to an industry, a company, its products / services, and last but not least, with the target audience. Each social network has different characteristics and, in addition, a different philosophy.The authors aim to carry out a bibliographic research in this paper to highlight the extent to which social networks are used. As a result, a top of social networks will be built to help raise awareness, promote products, and consolidate a strong customer-company relationship. The authors will also realize a statistical analysis of online social media networks to determine their key indicators, traffic on each platform, time spent by a user on that platform, and other key indicators, through an online tool. The results of this paper consist in presenting two classifications: the first from the perspective of the companies and the second from the perspective of social network users.

DynaMo: Dynamic Community Detection by Incrementally Maximizing Modularity

Community detection is of great importance for online social network analysis. The volume, variety and velocity of data generated by today's online social networks are advancing the way researchers analyze those networks. For instance, real-world networks, such as Facebook, LinkedIn and Twitter, are inherently growing rapidly and expanding aggressively over time. However, most of the studies so far have been focusing on detecting communities on the static networks. It is computationally expensive to directly employ a well-studied static algorithm repeatedly on the network snapshots of the dynamic networks. We propose DynaMo, a novel modularity-based dynamic community detection algorithm, aiming to detect communities of dynamic networks as effective as repeatedly applying static algorithms but in a more efficient way. DynaMo is an adaptive and incremental algorithm, which is designed for incrementally maximizing the modularity gain while updating the community structure of dynamic networks. In the experimental evaluation, a comprehensive comparison has been made among DynaMo, Louvain (static) and 5 other dynamic algorithms. Extensive experiments have been conducted on 6 real-world networks and 10,000 synthetic networks. Our results show that DynaMo outperforms all the other 5 dynamic algorithms in terms of the effectiveness, and is 2 to 5 times (by average) faster than Louvain algorithm.