The neighborhood role in the linear threshold rank on social networks

Riquelme, Fabián; González-Cantergiani, Pablo; Molinero, Xavier; Serna, María

doi:10.1016/j.physa.2019.121430

Cited by 10 publications

(4 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our experiments, we fixed the thresholds for the resistance to be the single majority criteria. Other thresholds were considered in [27]. It would of interest to analyze what determines the best threshold for a network.…”

Section: Discussionmentioning

confidence: 99%

Forward and backward linear threshold ranks

Blesa

García-Rodríguez

Serna

2021

Proceedings of the 2021 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining

Self Cite

View full text Add to dashboard Cite

Who are the most important or influencing actors of a social network? There are many ways to answer this question, but node centrality is probably the theoretical concept that better captures it and formalizes it. Traditional centrality measures, like the degree, closeness or betweenness of the actors, are purely based on topological properties of the network and that is not enough. Nowadays social networks are much more than simple topological structures, specially because they are very much involved in the spreading of information and have become a media themselves. Lately, models capturing that diffusion effect are enriching our theoretical knowledge about them. Examples of it are the Independent Cascade model and the Linear Threshold model and, based on them, a new generation of centrality measures are being proposed, e.g. the Independent Cascade Rank and the Linear Threshold Rank (LTR) [28].The focus of this research is on analyzing the effect of the initial activation set in the Linear Threshold Rank (LTR). The rank assigns to each actor the size of the set influenced by taking as initial activation set the actor and its immediate neighborhood. In the LTR both predecessors and successors are considered as immediate neighbors. We propose to analyze here the effect on the rank when the initial activation set contains only successors (Forward LTR, FwLTR), or only predecessors (Backward LTR, BwLTR). We perform an experimental analysis in a data set formed by a selection of networks with a variety of parameters. We compare our proposed measures among themselves and with other classical measures. Our results show that LTR and BwLTR behave quite similarly, while FwLTR is different. Furthermore the LTR variations define measures that are different from the other centrality measures considered in the study. To make the comparison, we use the same parameters and statistics used in [28] together with other taken from Social Sciences: the Gini coefficient and a categorization analysis. This second kind of new comparative analysis is of independent interest.

show abstract

Section: Discussionmentioning

confidence: 99%

Forward and backward linear threshold ranks

Blesa

García-Rodríguez

Serna

2021

Proceedings of the 2021 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining

Self Cite

View full text Add to dashboard Cite

show abstract

“…As a result, a width = 24 and a length = 14 was obtained. For the width, the following three optimal combinations were found (players are ordered according to Appendix A): Solution 1: {2, 5, 6, 7, 8, 9, 10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26, 27} Solution 2: {3, 5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27} Solution 3: {4,5,6,7,8,9,10,11,12,13,14,…”

Section: Methodsmentioning

confidence: 99%

“…It is known that any simple game can be represented by an influence game, and vice versa [8]. Furthermore, influence games and influence graphs have been already used to solve problems of multi-agent systems [8], social network analysis [9,10], and collective decision-making models [11,12].…”

Section: Introductionmentioning

confidence: 99%

Extremal Coalitions for Influence Games Through Swarm Intelligence-Based Methods

Riquelme¹,

Olivares²,

Mu駉z³

et al. 2022

Computers, Materials &Amp; Continua

Self Cite

View full text Add to dashboard Cite

An influence game is a simple game represented over an influence graph (i.e., a labeled, weighted graph) on which the influence spread phenomenon is exerted. Influence games allow applying different properties and parameters coming from cooperative game theory to the contexts of social network analysis, decision-systems, voting systems, and collective behavior. The exact calculation of several of these properties and parameters is computationally hard, even for a small number of players. Two examples of these parameters are the length and the width of a game. The length of a game is the size of its smaller winning coalition, while the width of a game is the size of its larger losing coalition. Both parameters are relevant to know the levels of difficulty in reaching agreements in collective decision-making systems. Despite the above, new bio-inspired metaheuristic algorithms have recently been developed to solve the NP-hard influence maximization problem in an efficient and approximate way, being able to find small winning coalitions that maximize the influence spread within an influence graph. In this article, we apply some variations of this solution to find extreme winning and losing coalitions, and thus efficient approximate solutions for the length and the width of influence games. As a case study, we consider two real social networks, one formed by the 58 members of the European Union Council under nice voting rules, and the other formed by the 705 members of the European Parliament, connected by political affinity. Results are promising and show that it is feasible to generate approximate solutions for the length and width parameters of influence games, in reduced solving time.

show abstract

“…Furthermore, a generalization of LTR measure was proposed in such a way the neighborhood level was changed by neighbors(i, ) = {j ∈ V | dist(i, j) ≤ }, where dist(i, j) denotes the length of the shortest undirected path between nodes i and j. It was also shown that using neighbors at distance larger than 2, the rankings distinguish better the influential actors [51].…”

Section: Centrality In Social Network Analysismentioning

confidence: 99%