Multi-relational Social Networks Research Articles

Lifelong and multirelational community detection to support social and collaborative e‐learning

AbstractLearners' community choice has a crucial role in e‐learning effectiveness. Indeed, individual and structural factors (i.e., learners pre‐existing profiles and networks of interactions) significantly affect how learners develop a collaborative e‐learning environment. In this context, social network analysis, singularly community detection has been a good approach to improve collaborative environment through discovering pertinent learners' communities and new effective relations. However, with social media emergence, many real‐world networks, such as learners' networks, evolve the connections in multiple layers, where each layer represents a different type of relationship. This, combined with their continuous evolution over time, has brought new challenges to the field of community detection. Thus, this paper proposes a new configurable algorithm for detecting collaborative and lifelong communities within dynamic multirelational social learners' networks. To do so, this algorithm is based on a graph model to represent these different interactions as well as different learners' profiles and characteristics. Moreover, by using particle swarm optimization, it aims to optimize a configurable combined metric to detect the most relevant community appropriate to a given situation. Finally, it considers the temporal dimension to find the final lifelong community. By the end of this paper, experimental results using synthetic networks prove that the proposed algorithm achieves better results compared with other community detection algorithms. Therewith, experiments on a real e‐learning network show this algorithm's role in improving collaboration within learners' network.

Reprint of: Using Trellis software to enhance high-quality large-scale network data collection in the field

Trellis is a mobile platform created by the Human Nature Lab at the Yale Institute for Network Science to collect high-quality, location-aware, off-line/online, multi-lingual, multi-relational social network and behavior data in hard-to-reach communities. Respondents use Trellis to identify their social contacts by name and photograph, a procedure especially useful in low-literacy populations or in contexts where names may be similar or confusing. We use social network data collected from 1,969 adult respondents in two villages in Kenya to demonstrate Trellis’ ability to provide unprecedented metadata to monitor and report on the data collection process including artifactual variability based on surveyors, time of day, or location.

Approaches to Assessing the Socio-Economic Consequences of the COVID-19 Pandemic Using Computer Simulation

In the context of the coronavirus pandemic, there is an increasing need to develop methods for scientifically based assessment of the consequences both at the level of the country's economy and at the regional level. One of the acute problems of the development of the Russian economy in the context of the coronavirus pandemic is the conflict between measures to protect the life and health of people and the fall in economic activity. To support the economy, countries are taking anti-crisis measures, which are aimed primarily at overcoming serious consequences in the most vulnerable sectors. As part of the study, to assess the socio-economic consequences of the epidemic and reproduce forecasts, modern simulation tools are used - agent-based modeling. Agent-based models allow you to use software of various classes, including neural networks, mathematical models, 3D-4D add-ons and other technologies that can visualize the results of scenario predictive estimates and computational experiments. The aim of the study is to develop methods and techniques for forecasting and scenario modeling of the socio-economic consequences of viral epidemics. For the study, a detailed statistical and analytical database was formed, adaptive blocks were developed with the possibility of additional inclusion of indicators. The software implementation included three functional blocks: demographic, economic and epidemiological, as well as three categories of agents within each subject of the Russian Federation with individual characteristics based on accepted world practice. The software tool chosen to implement the research objectives is the platform for creating agent-based models "AnyLogic". The study was carried out on the example of the following subjects of the Russian Federation: Murmansk region, Krasnodar region, Sverdlovsk, Samara and Voronezh regions. Based on the results of the study, an architecture of an agent-based model was developed, which makes it possible to evaluate restrictive measures and regulations in terms of the socio-economic consequences of a pandemic. As a result of the study, methods and algorithms for agent-based modeling of the socio-economic consequences of viral epidemics were developed, taking into account spatial and communicative interactions. To fulfill the objectives of the study, at the first stage, an analysis of scientific methods for forecasting and building various models for assessing the consequences of macroeconomic decisions and models for the spread of viral epidemics was carried out. At the second stage, an agent-based model was developed, which took into account structured and unstructured information, including the socio-demographic and economic characteristics of the regions, such as morbidity and mortality, employment rates, as well as measures taken by the regions to counter the spread of COVID-19. In terms of social interaction between agents, the study implemented a dynamic multi-relational (MRN) social network of agents, the structure of which changes during the introduction of quarantine measures that limit the degree of interaction between them. The introduction of different specific values of individual characteristics within a population of agents of the same type makes it possible to assess the socio-economic consequences of viral epidemics with the maximum degree of detail - at the level of individuals. Further development of this area of research will include refinement of the developed model for analyzing the consequences of the spread of viral epidemics in terms of the socio-economic development of territorial systems based on the obtained forecast scenarios.

Using Trellis software to enhance high-quality large-scale network data collection in the field

Trellis is a mobile platform created by the Human Nature Lab at the Yale Institute for Network Science to collect high-quality, location-aware, off-line/online, multi-lingual, multi-relational social network and behavior data in hard-to-reach communities. Respondents use Trellis to identify their social contacts by name and photograph, a procedure especially useful in low-literacy populations or in contexts where names may be similar or confusing. We use social network data collected from 1,969 adult respondents in two villages in Kenya to demonstrate Trellis’ ability to provide unprecedented metadata to monitor and report on the data collection process including artifactual variability based on surveyors, time of day, or location.

Leveraging multi-level dependency of relational sequences for social spammer detection

Much recent research has shed light on developing the relation-dependent but the content-independent framework for social spammer detection. This is mainly because the relation among users is difficult to be altered when spammers attempt to conceal their malicious intentions. Our study investigates the spammer detection problem in the context of multi-relation social networks and makes an attempt to fully exploit the sequences of heterogeneous relations for enhancing the detection accuracy. Specifically, we present the Multi-level Dependency Model (MDM). The MDM is able to exploit the user’s long-term dependency hidden in their relational sequences along with short-term dependency. Moreover, MDM fully considers short-term relational sequences from the perspectives of individual-level and union-level, due to the fact that the type of short-term sequences is multi-folds. Experimental results on a real-world multi-relational social network demonstrate the effectiveness of our proposed MDM on multi-relational social spammer detection.

Community detection in multi-relational social networks based on relational concept analysis

Multi-relational community discovering in heterogeneous social networks is an important issue. Many approaches has been proposed for community discovering in heterogeneous networks. However, they have focused only on topological properties of these networks, ignoring the embedded semantic information. As the solution to this information glut limit, we propose, in this paper, a new multi-relational community discovering approach which incorporates the multiple types of objects and relationships, derived from heterogeneous networks. Firstly, we propose to construct the Concept Lattice Family CLF to represent the different objects and relations of the heterogeneous social networks based on the relational concept analysis techniques. Then after we introduce a new algorithm that explores such CLF and extract the multi-relational communities. Carried out experiments on real-datasets enhance the effectiveness of our proposal and open promising issues.

Detect potential relations by link prediction in multi-relational social networks

Potential relation detecting on social network has become more important for decision making in many business disciplines, such as marketing, business strategy, human resources development, finance planning, business transformation, insurance policy design, and tourism management. People are used to seeking useful information from the relationships among social members to support their decisions on investment, partner seeking and marketing. Corporations are seeking opportunities to leverage them for “word of mouth” advertising based on the relations between the customers. When we collect and observe relationships between people, missing or redundant relations unavoidably occur since the time and cost restrictions in market or social investigation prevent us to discover all the relations. Moreover, since the social relations are changing constantly, current social relations may disappear, and new relations will be established. Many trade and social networks consist of multiple types of relations between the individuals. This paper presents an efficient method to detect the potential and future social relations between individuals in multi-relational social networks using link prediction. First, we calculate the belief of each individual by belief propagation on each type of relations. Based on the belief vectors, the similarities between various types of relations are computed to measure their mutual influence. Based on the similarities between various types of relations, we model link prediction as the problem of matrix completion by optimizing its max-norm constrained formulation. We propose a projected gradient descent optimization algorithm which is scalable to large size networks. Empirical results on real multi-relational social networks demonstrate that the predicting results of our algorithm have higher quality compared with other similar algorithms.

A Network Autocorrelation Model to Predict Repeat Purchases in Multi-Relational Social Networks: Evidence from Online Games

It is well established in the literature that customer retention is at least as important as customer acquisition, especially in the freemium-based virtual economy, in which individuals are connected by multi-relational social networks, such as the “friendships” and “guilds” (virtual teams) functions in the context of online games. Yet, we do not know how these multi-relational social network effects simultaneously affect repeat purchases. In this study, we seek to examine the interdependent nature of repeat purchases of online game players embedded in multi-relational social networks. We examine two types of social network effects in the “friendship” function: a reinforcement (positive) effect and an attenuation (negative) effect. We also examine the herding effect (behavioral conformity within a virtual team) in the “guild” function. We develop a new hierarchical Bayesian model that supports a response variable for repeat purchases and that follows a Poisson distribution, and we simultaneously include both the two social network effects (reinforcement and attenuation) from the “friendship” function and also the moderating role of the herding effect from the “guild” function. Our study shows that both the reinforcement and attenuation social network effects have a significant effect on repeat purchases; however, the direction of the behavioral correlation of these two social network effects is starkly distinct: the reinforcement effect leads to repeat purchases among direct friends; in contrast, the attenuation effect leads to a divergence of repeat purchases among players who are indirect friends but share common friends. Moreover, we find that the size of the reinforcement effect is attenuated when the herding effect is accounted for. Besides theoretically developing and empirically validating a new method to predict repeat purchases in multi-relational social networks, this study has managerial implications for online game providers to develop effective targeting strategies to encourage players to engage in repeat purchases by leveraging multi-relational social network effects. Broader theoretical and practical implications for encouraging repeat purchases by leveraging social network effects in other online platforms are also discussed.

Using triangles and latent factor cosine similarity prior to improve community detection in multi‐relational social networks

SummaryCommunity detection is a key to understanding the structure of complex networks. Communities, or clusters, are groups of vertices having higher probability of being connected to each other than to the members in other groups. Considering the importance of triangle structures, we first propose σ‐tensor to model ordinary relationships and triangle relationships simultaneously. Then, we propose a simple but effective latent factor prior, ie, latent factor cosine similarity prior, to improve community detection. The latent factor cosine similarity prior is a kind of statistics of the well‐defined synthetic multi‐relational social networks. It is based on a key observation that most latent feature factors of intra‐group members in these networks are highly similar according to cosine similarity measure. Using this prior along with the RESCAL tensor factorization model, we can obtain a superior latent feature factor matrix. Moreover, N‐RESCAL model, a variant of RESCAL model, and its corresponding algorithm N‐RESCAL‐ALS are proposed for the simplicity and the removal of the limit of cosine similarity. Once the latent factor matrix is obtained by factorizing σ‐tensor using N‐RESCAL model, we apply agglomerative clustering algorithm for community discovery. We call this framework as TNRA. Experiment results on several real‐world datasets are surprisingly promising, clearly demonstrating the power of the proposed prior and the effectiveness of our proposed methods.

Multirelational Social Recommendations via Multigraph Ranking.

Recommender systems aim to identify relevant items for particular users in large-scale online applications. The historical rating data of users is a valuable input resource for many recommendation models such as collaborative filtering (CF), but these models are known to suffer from the rating sparsity problem when the users or items under consideration have insufficient rating records. With the continued growth of online social networks, the increased user-to-user relationships are reported to be helpful and can alleviate the CF rating sparsity problem. Although researchers have developed a range of social network-based recommender systems, there is no unified model to handle multirelational social networks. To address this challenge, this paper represents different user relationships in a multigraph and develops a multigraph ranking model to identify and recommend the nearest neighbors of particular users in high-order environments. We conduct empirical experiments on two real-world datasets: 1) Epinions and 2) Last.fm, and the comprehensive comparison with other approaches demonstrates that our model improves recommendation performance in terms of both recommendation coverage and accuracy, especially when the rating data are sparse.

A Multirelational Social Network Analysis of an Online Health Community for Smoking Cessation.

BackgroundOnline health communities (OHCs) provide a convenient and commonly used way for people to connect around shared health experiences, exchange information, and receive social support. Users often interact with peers via multiple communication methods, forming a multirelational social network. Use of OHCs is common among smokers, but to date, there have been no studies on users’ online interactions via different means of online communications and how such interactions are related to smoking cessation. Such information can be retrieved in multirelational social networks and could be useful in the design and management of OHCs.ObjectiveTo examine the social network structure of an OHC for smoking cessation using a multirelational approach, and to explore links between subnetwork position (ie, centrality) and smoking abstinence.MethodsWe used NetworkX to construct 4 subnetworks based on users’ interactions via blogs, group discussions, message boards, and private messages. We illustrated topological properties of each subnetwork, including its degree distribution, density, and connectedness, and compared similarities among these subnetworks by correlating node centrality and measuring edge overlap. We also investigated coevolution dynamics of this multirelational network by analyzing tie formation sequences across subnetworks. In a subset of users who participated in a randomized, smoking cessation treatment trial, we conducted user profiling based on users’ centralities in the 4 subnetworks and identified user groups using clustering techniques. We further examined 30-day smoking abstinence at 3 months postenrollment in relation to users’ centralities in the 4 subnetworks.ResultsThe 4 subnetworks have different topological characteristics, with message board having the most nodes (36,536) and group discussion having the highest network density (4.35×10−3). Blog and message board subnetworks had the most similar structures with an in-degree correlation of .45, out-degree correlation of .55, and Jaccard coefficient of .23 for edge overlap. A new tie in the group discussion subnetwork had the lowest probability of triggering subsequent ties among the same two users in other subnetworks: 6.33% (54,142/855,893) for 2-tie sequences and 2.13% (18,207/855,893) for 3-tie sequences. Users’ centralities varied across the 4 subnetworks. Among a subset of users enrolled in a randomized trial, those with higher centralities across subnetworks generally had higher abstinence rates, although high centrality in the group discussion subnetwork was not associated with higher abstinence rates.ConclusionsA multirelational approach revealed insights that could not be obtained by analyzing the aggregated network alone, such as the ineffectiveness of group discussions in triggering social ties of other types, the advantage of blogs, message boards, and private messages in leading to subsequent social ties of other types, and the weak connection between one’s centrality in the group discussion subnetwork and smoking abstinence. These insights have implications for the design and management of online social networks for smoking cessation.

Algorithm for Discovering Community in Multi-Relational Social Network Based on Modified Common Neighbors Similarity

The multi-relational social network is a new social network, which is proposed in recent years and more complex than single relational social network. Therefore, the existing community discovery algorithm is lack and their results are crude. The paper proposes a new algorithm for discovering community in multi-relational social network based on modified common neighbors similarity. The core idea is the more common neighbors are shared between two nodes, the more similar the nodes are. Finally, we test our generalized algorithm on an artificial sparse network in four dimensions.

A New Framework for Churners’ Influence Analysis Using Call Data Records

Predicting customer churn is of paramount importance in telecommunication companies. The taxonomy of churn reports that, not only individual constraints but also social factors can create users’ propensity to churn. This study uses real world call data records (CDR) to extract the social relationships among mobile phone users and build a multi relational social network, where the influence of users diffuses. The research is conducted to propose a framework that enhances the actionable value of social influence of predicted churners and to examine the parameters that control the churn information diffusion in the telecommunication networks.

A stochastic agent-based model of pathogen propagation in dynamic multi-relational social networks.

We describe a general framework for modeling and stochastic simulation of epidemics in realistic dynamic social networks, which incorporates heterogeneity in the types of individuals, types of interconnecting risk-bearing relationships, and types of pathogens transmitted across them. Dynamism is supported through arrival and departure processes, continuous restructuring of risk relationships, and changes to pathogen infectiousness, as mandated by natural history; dynamism is regulated through constraints on the local agency of individual nodes and their risk behaviors, while simulation trajectories are validated using system-wide metrics. To illustrate its utility, we present a case study that applies the proposed framework towards a simulation of HIV in artificial networks of intravenous drug users (IDUs) modeled using data collected in the Social Factors for HIV Risk survey.

Multi-relational social networks in a large-scale MMORPG

We analyze multi-relational social interaction networks in a large-scale commercial Massively Multiplayer Online Role-Playing Game(MMORPG). Our work is based on data from AION, currently the world's second most-played MMORPG with 3.4 million subscribers as of mid 2010, created and serviced by NCSoft, Inc. We construct and characterize six distinct interactivity networks (Friend, Private Messaging, Party invitation, Trade, Mail, and Shop), each representing diverse player interaction types.

Statistical Evaluation of Algebraic Constraints for Social Networks

A multirelational social network on a set of individuals may be represented as a collection of binary relations. Compound relations constructed from this collection represent various labeled paths linking individuals in the network. Since many models of interest for social networks can be formulated in terms of orderings among these labeled paths, we consider the problem of evaluating an hypothesized set of orderings, termed algebraic constraints. Each constraint takes the form of an hypothesized inclusion relation for a pair of labeled paths. In this paper, we establish conditions under which sets of such constraints may be regarded as partial algebras. We describe the structure of constraint sets and show that each corresponds to a subset of consistent relation bundles between pairs of individuals. We thereby construct measures of fit for a given constraint set. Then, we show how, in combination with the assumption of various conditional uniform multigraph distributions, these measures lead to a flexible approach to the evaluation of fit of an hypothesized constraint set. Several applications are presented and some possible extensions of the approach are briefly discussed.

Blockmodels: Interpretation and evaluation

Abstract Many methods for the description of social network structural properties are concerned with the dual notions of social position and social role. Common goals of these methods are to represent patterns in complex social network data in simplified form, to reveal sets of actors who are similarly embedded in networks of relations, and to describe the associations among relations in multirelational social networks. Often these representations take the form of a blockmodel. In a blockmodel actors are assigned to positions and network relations are presented among positions, rather than among actors. The literature on blockmodels is extensive and is overflowing with computation and applications of blockmodels. However, there is a surprising lack of attention to two very important aspects of blockmodel analyses: the interpretation and evaluation of the results. The purpose of this paper is to focus on these topics, primarily reviewing and synthesizing the approaches to interpretation and evaluation currently in use.