Influence maximization on signed networks under independent cascade model

Liu, Wei; Chen, Xin; Chen, Ling; Chen, Bolun

doi:10.1007/s10489-018-1303-2

Cited by 37 publications

(12 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This algorithm uses the linear threshold model and its extended LT-N to maximize the influence of cumulative features by considering positive and negative relationships. Liu et al [31] used an independent cascading model in signed networks to study influence maximization and proposed a greedy algorithm based on independent paths to maximize the spread of positive influence in signed networks. Li et al [32][33][34][35][36] investigated Community-diversified Influence Maximization (CDIM) problem to find k nodes and also proposed a metric to measure the community-diversified influence and addresses a series of computational challenges.…”

Section: Related Workmentioning

confidence: 99%

Influence Maximization in Signed Networks by Enhancing the Negative Influence

Dai

2021

IEEE Access

View full text Add to dashboard Cite

With the advancement of science and technology, research on influence maximization in networks has become a hot spot. In social networks, there is not only a positive relationship among nodes but also a negative impact, and the negative impact often plays a greater role than the positive impact, as observed for shopping websites or online votes. This paper proposes a method based on an independent cascade model by emphasizing the negative impact in symbolic networks to solve the problem of influence maximization. First, an algorithm that is based on an independent path and that emphasizes negative influence is designed to obtain the probability among nodes. Based on the activation probability, an algorithm is proposed to identify nodes that could have the greatest impact on the influence increment from the seeds. Finally, the seed set is confirmed based on the influence in the corresponding symbol network. In the experiment performed on real-world network data, the result indicate that the proposed algorithm causes more substantial influence propagation than do other algorithms.

show abstract

Section: Related Workmentioning

confidence: 99%

Influence Maximization in Signed Networks by Enhancing the Negative Influence

Dai

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…We generate the instances G t of the social network by using dynamic link prediction method [55], [56] at the beginning of every time slot or hop t. Again, the selective link prediction effectively reduces computational complexity, and hence, we predict links for only necessary newly infected nodes instead of all the nodes in the whole network.…”

Section: Dynamic Link Predictionmentioning

confidence: 99%

CM2D: Cost Minimization Under the Competitive Market in Dynamic Social Networks

Talukder

Hong

2020

IEEE Access

View full text Add to dashboard Cite

The Reverse Influence Maximization (RIM) model deals with the viral marketing cost minimization in social networks. On the other hand, the Influence maximization (IM) technique finds the small number of influential users that maximize the viral marketing profit. Here, the profit is defined by the maximum number of nodes that can be influenced by seed users when they are initially activated. On the other hand, the cost is measured by the minimum number of nodes required to activate all the seed nodes. However, most of the existing studies focus on profit maximization without considering the seeding cost. Moreover, the most profitable node may not always be a cost-effective one. Thus, in this research, we introduce an epidemic Cost Minimization model under the Competitive Market in Dynamic social networks (CMCMD, i.e., CM2D). In the CM2D model, we modify the Susceptible-Infected-Recovered (SIR) technique as the Reverse SIR (RSIR) model to employ in the node activation process. The proposed CM2D model also uses the greedy Set Cover approximation technique to optimize the seeding cost. Moreover, the CM2D model tackles the challenging issues of the RIM problem more efficiently than the existing models. Finally, we simulate our model using datasets of two synthesized and four real datasets, and the results show that the CM2D model outperforms the existing models.

show abstract

“…Liu et. al., have developed a greedy algorithm for the IM problem [15]. It constructs the set of spreading paths.…”

Section: Greedy Approaches (Aç Gözlü Yaklaşımlar)mentioning

confidence: 99%

A New Greedy Algorithm For Influence Maximization On Signed Social Networks

Şimşek¹

2019

GMBD

View full text Add to dashboard Cite

Sosyal etki insanların görüşlerini şekillendiren büyük olgulardan biridir. Bu bakımdan, Etki Maksimizasyonu (EM) problemi viral pazarlama, kamuoyu şekillendirme gibi pratik faydaları olduğu için sosyal ağ analizinde en fazla ilgili çeken araştırma alanlarından biridir. EM probleminin amacı bir sosyal ağ üzerindeki etkili kişi olarak adlandırılan az sayıdaki kişiyi kullanarak bir etkinin (bir fikir veya reklam) ağ üzerindeki yayılımını maksimize etmektir. Etkili kişilerin tespiti birçok durumda NP-Zor bir olasılıksal en iyileme problemidir. Bundan dolayı, EM problemi için birçok algoritma geliştirilmiştir ve geliştirilmeye devam etmektedir. Ne var ki, geliştirilen algoritmalar henüz çözüm kalitesi ve hız açısından istenen seviyede değildirler. Bu çalışmada, bireyler arasındaki olumlu ve olumsuz ilişkileri göz önünde bulunduran işaretli EM problemine odaklanılmıştır. Bu amaçla, en iyi k adet etkili kişiyi tespit etmek için Elitist Aç Gözlü Algoritma (EGA) olarak adlandırılan bir aç gözlü algoritma geliştirmiştir. EGA'nın performansı 2 adet açık veriseti üzerinde rasgele seçim, çıkış derecesi merkeziliği, ve bir güncel algoritma ile kıyaslanmıştır. EGA çözüm kalitesi açısından rakiplerine göre daha iyi sonuçlar vermiştir.https://dx.The social influence is one of the major phenomenons that shape people's decisions. In this respect, Influence Maximization (IM) problem is one of the most attractive research topics in the social network analysis because its practical benefits in viral marketing, public opinions shaping etc. The IM problem aims to maximize the spread of an influence (e.g. an opinion, an advertisement) in a social network by using a small number of the most effective individuals, whom is called influencers. Detecting the influencers is the NP-Hard combinatorial optimization problem in most cases. Therefore, many algorithms have been and are being developed for the IM problem. However, the algorithms have not yet achieved to the desired solution quality and speed. In this study, we focused on the signed IM problem that considers both positive and negative influence between the individuals. For this purpose, we developed a greedy algorithm called the Elitist Greedy Algorithm (EGA) for detecting top-k influencers set. We compared the EGA's performance on 2 public datasets with random seed selection, out degree heuristic, and one stateof-the-art greedy algorithm. The EGA outperforms the competitors in terms of the achieved total influence. https://dx.

show abstract

Influence maximization on signed networks under independent cascade model

Cited by 37 publications

References 46 publications

Influence Maximization in Signed Networks by Enhancing the Negative Influence

Influence Maximization in Signed Networks by Enhancing the Negative Influence

CM2D: Cost Minimization Under the Competitive Market in Dynamic Social Networks

A New Greedy Algorithm For Influence Maximization On Signed Social Networks

Contact Info

Product

Resources

About