Title: Mathematical modelling of abnormal account detection on social media platform based on improved edge weight
Authors: Yu Han
Addresses: Department of Basic Courses, Shaanxi Open University, Xi'an 710119, China
Abstract: In order to overcome the problems of traditional methods such as large time consumption, high missed detection rate and low recall rate of detection results, a mathematical modelling method of abnormal account detection on social media platform based on improved edge weight was proposed. The social media platform is regarded as a directed social graph, the node states in the social graph are judged, and the edge weight of the social graph is used to calculate the edge potential. First, take the improved edge weight processing results as the operation constraints of the model. Then, build the abnormal account detection mathematical model. Finally, input the user account information into the model, and output the abnormal account detection results. The experimental results show that the maximum detection time consumption of the proposed method is only 0.88 min, the highest missed detection rate is only 2.9%, and the lowest recall rate can reach 98.49%.
Keywords: social graph; edge weights; node status; abnormal account; anomaly detection.
DOI: 10.1504/IJWBC.2023.131400
International Journal of Web Based Communities, 2023 Vol.19 No.2/3, pp.187 - 197
Received: 31 Oct 2021
Accepted: 21 Feb 2022
Published online: 09 Jun 2023 *