Effective method prevents spread of false information on social networks
The group's goal is to study the mechanisms and patterns of information dissemination in general and misinformation in particular on online social networks; identify and analyze the characteristics of misinformation, factors on social networks that play an important role in disseminating information; at the same time, propose effective methods and techniques to help limit the spread of false information; focus on the problem in which information is distributed from many sources, on many different topics, the distribution mechanism changes over time, and at the same time must satisfy the constraints on the cost of prevention. The team also builds, tests, and evaluates the effectiveness of the proposed methods on simulated datasets and data from real social networks.
There have been many proposed information transmission models in the world. These models have played a role in theoretical research as well as practice on information transmission with many different applications. In which, the three most widely applied propagation models are: Linear Threshold model, discrete random model in which information is propagated in discrete time steps; Level-independence model with the main feature being the independent propagation of information by each edge; and SIR model, a basic mathematical model of epidemics, was introduced in the classic paper of Kermack and McKendrick, and then also applied in forecasting information transmission. This topic has used, inherited and improved the Linear Threshold model and the Degree of Independence model in the research contents.
Limiting the spread of false information on social networks is done on the basis of detecting information propagation paths (propagation mechanisms, propagation diagrams), then identifying intermediate network nodes that play an important role in spreading to conduct prevention. There are many techniques to limit the spread of false information, including three outstanding techniques: Influence prevention techniques, Disabling users or link sets, and Information decontamination techniques.
Figure 1. Overall model of the system
Figure 2. System interface
In this topic, the research team followed the technique of disabling users or linking sets, which means that on the network there will be an isolated set of users, false information propagates before this set of nodes and stops, cannot propagate further to other nodes. This set of network nodes is considered a barrier, preventing the spread of false information. Specifically, the team used the "Vaccination" method, with 2 algorithms: IGA Improved Greed Algorithm and GEA Extended Greed Algorithm to prevent the spread of false information on online social networks linear with the test simulation dataset. Both algorithms are described in a scientific report on methods and techniques to prevent the spread of false information on online social networks.
The project built a pilot program to prevent the spread of false information on online social networks. The team conducted a test and evaluated the results of running a program to prevent the spread of false information on online social networks with the input of real social networks that are filtered for privacy factors.
In addition, the team published 02 scientific articles related to the research content: 01 international paper published in the journal Optimazation letter under the list of SCIE in 2022 (SCIE Q1); 01 article in the Journal of Information and Communication Technology research, development and application, volume 2021, number 2, pages 104-111, 2021.
Translated by Phuong Huyen
Link to Vietnamese version