Ce diaporama a bien été signalé.
Nous utilisons votre profil LinkedIn et vos données d’activité pour vous proposer des publicités personnalisées et pertinentes. Vous pouvez changer vos préférences de publicités à tout moment.

Information dissemination in scale-free networks

26 vues

Publié le

The study of information dissemination in social networks is of particular importance in many areas as marketing, politics and security for example. Various strategies are being developed to disseminate information, those aimed at disseminating information widely and those aimed at disseminating information in a more confidential manner to make it scarce. In this paper, we adapt a model dedicated to spreading rumours by word of mouth in a physical space to the context of social networks. We compare two modes of dissemination based on profusion or scarcity and study the impact of the choice of the initial node. The results obtained show to what extent each mode exploits the social network topology and especially the influence of hubs.

Publié dans : Données & analyses
  • Soyez le premier à commenter

  • Soyez le premier à aimer ceci

Information dissemination in scale-free networks

  1. 1. INFORMATION DISSEMINATION IN SCALE-FREE NETWORKS: PROFUSION VS SCARCITY Laurent Brisson, Philippe Collard, Martine Collard, Erick Stattner
  2. 2. Why Many applications: marketing crowfunding security ... What Information could be rumors, ideas, sentiments, desires ... We are not interested in the content but the way the information propagates Where Social networks Scale-free network generated with Barab�si Albert algorithm How Information can spread if: it is popular (diffusion by profusion) it remains confidential (diffusion by scarcity) INFORMATION DISSEMINATION IN SCALE-FREE NETWORKS: PROFUSION VS SCARCITY Laurent Brisson, Philippe Collard, Martine Collard, Erick Stattner The 6th International Conference on Complex Networks and Their Applications - November 29 / December 01 2017 SIMULATING INFORMATION DISSEMINATION Context 2
  3. 3. What is the impact of the ... dissemination mode? first informed node? [on] The end the number of people finally informed the time at the end of the diffusion [on] The dynamics the maximum number of people simultaneously informed the time of this diffusion peak [on] The replicability the mean standard deviation of results given a set of parameters INFORMATION DISSEMINATION IN SCALE-FREE NETWORKS: PROFUSION VS SCARCITY Laurent Brisson, Philippe Collard, Martine Collard, Erick Stattner The 6th International Conference on Complex Networks and Their Applications - November 29 / December 01 2017 SIMULATING INFORMATION DISSEMINATION Problem 3
  4. 4. ODS model probabilistic model based on the foundations of SIR and DK models receiving agents decide whether or not to transmit the information 3 states / compartments Open-minded [O]: no access to the information Disseminator [D]: currently disseminating Stifler [S]: had access but no longer disseminating 2 transitions OtoD: for each Open-minded agent in the neighborhood of a Disseminator the Open-minded agent decides whether or not he becomes a Disseminator depending on profusion or scarcity mode DtoS: a Disseminator remains in its state for a time depending on the global DPeriod parameter INFORMATION DISSEMINATION IN SCALE-FREE NETWORKS: PROFUSION VS SCARCITY Laurent Brisson, Philippe Collard, Martine Collard, Erick Stattner The 6th International Conference on Complex Networks and Their Applications - November 29 / December 01 2017 DISSEMINATION MODEL A compartment model 4
  5. 5. over Barab�si-Albert graph with 1,000 nodes (degree distribution follows a power law) with Experimental conditions First informed node Profusion vs Scarcity DPeriod 10 runs for each set of parameters 200,000 experiments performed INFORMATION DISSEMINATION IN SCALE-FREE NETWORKS: PROFUSION VS SCARCITY Laurent Brisson, Philippe Collard, Martine Collard, Erick Stattner The 6th International Conference on Complex Networks and Their Applications - November 29 / December 01 2017 EXPERIMENTAL RESULTS Agent-based simulation 5
  6. 6. Profusion Occurence Rate increases almost linearly with DPeriod Scarcity For a given Population Ratio there is a DPeriod threshold above which Occurence Rate is 1 Occurrence rate where dissemination reached a given PopulationRatio INFORMATION DISSEMINATION IN SCALE-FREE NETWORKS: PROFUSION VS SCARCITY Laurent Brisson, Philippe Collard, Martine Collard, Erick Stattner The 6th International Conference on Complex Networks and Their Applications - November 29 / December 01 2017 EXPERIMENTAL RESULTS Transmissibility potential 6
  7. 7. Profusion Time and intensity of diffusion peak increases with DPeriod Scarcity Time and intensity of diffusion peak remain unchanged Incidence curves according DPeriod INFORMATION DISSEMINATION IN SCALE-FREE NETWORKS: PROFUSION VS SCARCITY Laurent Brisson, Philippe Collard, Martine Collard, Erick Stattner The 6th International Conference on Complex Networks and Their Applications - November 29 / December 01 2017 EXPERIMENTAL RESULTS Incidence curve 7
  8. 8. Profusion Standard deviation increases with DPeriod Scarcity Standard deviation is low and stable Box Plot of Standard deviation of Number of Stiflers at convergence according DPeriod INFORMATION DISSEMINATION IN SCALE-FREE NETWORKS: PROFUSION VS SCARCITY Laurent Brisson, Philippe Collard, Martine Collard, Erick Stattner The 6th International Conference on Complex Networks and Their Applications - November 29 / December 01 2017 EXPERIMENTAL RESULTS Diffusion variability 8
  9. 9. Evolution of the mean degree of Disseminators over time Influence of first informed node Profusion Betweenness, Pagerank and Degree of first informed node are correlated with: number of Disseminators at dissemination peak number of Stiflers at convergence Scarcity Closeness of first informed node is correlated with: time of dissemination peak time at convergence INFORMATION DISSEMINATION IN SCALE-FREE NETWORKS: PROFUSION VS SCARCITY Laurent Brisson, Philippe Collard, Martine Collard, Erick Stattner The 6th International Conference on Complex Networks and Their Applications - November 29 / December 01 2017 EXPERIMENTAL RESULTS Influence of network topology 9
  10. 10. Agent-based simulations over a scale-free network Profusion: Most people could be reach if dissemination last enough ... ... but with a lot of uncertainty First informed node influence the 'intensity' Scarcity: Quicker dissemination Reproductible results First informed node influence the 'time' Perspectives Analyze the influence of various network topologies according to the diffusion mode Investigate the link between the nature of the message and the type of propagation on real datasets INFORMATION DISSEMINATION IN SCALE-FREE NETWORKS: PROFUSION VS SCARCITY Laurent Brisson, Philippe Collard, Martine Collard, Erick Stattner The 6th International Conference on Complex Networks and Their Applications - November 29 / December 01 2017 CONCLUSION 10

×