A mitigation strategy for the prevention of cascading trust failures in social networks

Abstract In the past decade, we have seen a massive growth in social networks and a day to day increase in their users. Loose constraints and almost no limitation in the propagation of information in these networks have resulted in a lot of false information, spam and inappropriate messages being exchanged among users. This has caused the creation of a trust crisis in which trust relationships turn into distrust. But, the real threat to the general integrity of social networks occurs when the removed trust edges in the social network result in more and more connections to turn into distrust and thus be removed from the corresponding trust network. This phenomenon is called cascading trust failure. The primary purpose of this research is to build upon the dynamic modeling of cascading failures due to the occurrence of a trust crisis within a unidirectional or bidirectional social network. After the model is created, the next step would be on the detection of neighboring edges that can be influenced by the trust crisis and may be removed because of it. The second purpose of the research is to propose a mitigation strategy for the prevention of cascading failures in trust relationships. This will be beneficial for the honest and free expression of opinions and experiences without their privacy getting compromised or the trust relationships being affected. The proposed model has four steps: (1) trust calculation and evaluation, (2) propagation, (3) updating and (4) the filtration process. In the model, important parameters such as changes in topology, cascading times and failure as well as the connectivity ratio are considered. The impact of these parameters is also investigated in the Facebook’s sparse network and Twitter’s ripple network. The performed evaluations show that the strength of trust relationships not only depend on profile similarity measures but also on the subjective, propagative, dynamic, event sensitive and the asymmetrical characteristics of trust. Also, it is shown that the sensitivity of users toward the change in their network topology is also a considerable factor in the occurrence and subsequent prevention of cascading trust failures. Based on the performed evaluations, the proposed mitigation strategy is capable of maintaining 95% and 92% of the trust relations when a trust crisis targets the highest trust values in the Facebook and Twitter’s networks respectively. This is a considerable increase from the 52% and 2% remaining edges in the occurrence of trust crisis in the previously proposed approaches. Also, the proposed approach has an average of 50% increase in reducing the number of cascading failures as well as their impact on the network.

[1]  Iraklis Varlamis,et al.  A Trust-Aware System for Personalized User Recommendations in Social Networks , 2014, IEEE Transactions on Systems, Man, and Cybernetics: Systems.

[2]  Mohammad Abdollahi Azgomi,et al.  Contextuality, Incompatibility and Biased Inference in a Quantum-like formulation of Computational Trust , 2014, Adv. Complex Syst..

[3]  Mohammad Khansari,et al.  Centrality measures for immunization of weighted networks , 2016 .

[4]  Fei Tan,et al.  Cascading failures of loads in interconnected networks under intentional attack , 2013 .

[5]  Jiliang Tang.  Computing Distrust in Social Media , 2015 .

[6]  Xue Yibo,et al.  Mitigation strategy against cascading failures on social networks , 2014, China Communications.

[7]  Akira Namatame,et al.  Improve Network's Robustness Against Cascade with Rewiring , 2013 .

[8]  Chen Hong,et al.  Improving the network robustness against cascading failures by adding links , 2013 .

[9]  Cécile Paris,et al.  A survey of trust in social networks , 2013, CSUR.

[10]  Matjaz Perc,et al.  Information cascades in complex networks , 2017, J. Complex Networks.

[11]  Kanliang Wang,et al.  A trust model for multimedia social networks , 2012, Social Network Analysis and Mining.

[12]  Jennifer Golbeck,et al.  Analyzing the Social Web , 2013 .

[13]  Marco Conti,et al.  The structure of online social networks mirrors those in the offline world , 2015, Soc. Networks.

[14]  Enrico Zio,et al.  Optimization of Cascade‐Resilient Electrical Infrastructures and its Validation by Power Flow Modeling , 2015, Risk analysis : an official publication of the Society for Risk Analysis.

[15]  Michalis Vazirgiannis,et al.  Vulnerability assessment in social networks under cascade-based node departures , 2015 .

[16]  Yan Jia,et al.  Identifying Vulnerable Nodes of Complex Networks in Cascading Failures Induced by Node-Based Attacks , 2013 .

[17]  Jie Wu,et al.  Generating trusted graphs for trust evaluation in online social networks , 2014, Future Gener. Comput. Syst..

[18]  Jingchi Jiang,et al.  Modeling cascading failures with the crisis of trust in social networks , 2015 .

[19]  Chen Jiang,et al.  Robustness of interdependent networks with different link patterns against cascading failures , 2014 .

[20]  Jianwei Wang,et al.  Mitigation strategies on scale-free networks against cascading failures , 2013 .

[21]  Frances M. T. Brazier,et al.  A robustness metric for cascading failures by targeted attacks in power networks , 2013, 2013 10th IEEE INTERNATIONAL CONFERENCE ON NETWORKING, SENSING AND CONTROL (ICNSC).

[22]  Kimmo Kaski,et al.  Calling Dunbar's numbers , 2016, Soc. Networks.