Community detection across multiple social networks based on overlapping users

Zhu, Ziqing; Zhou, Tao; Jia, Chenghao; Liu, Weijia; Liu, Bo; Cao, Jiuxin

doi:10.1002/ett.3928

Cited by 10 publications

(4 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With the wide application of Internet technology in the world, the number of users/terminals of all kinds has already exceeded one billion, and all fields have entered the informationization era to a certain extent [1][2][3]. However, as computer network technology continues to develop, the user scale continues to expand at the same time, accompanied by all kinds of user/terminal abnormal behavior affecting the normal operation of the network and the increasingly heavy network operation and maintenance, analysis work [4][5].…”

Section: Introductionmentioning

confidence: 99%

Abnormal User Behavior Detection for Industry Big Data Analysis

Wang,

2024

Applied Mathematics and Nonlinear Sciences

View full text Add to dashboard Cite

The security of user information and the precision of user services are paramount, necessitating effective detection of abnormal user behavior in vast datasets. This study introduces the QGAN-BDE algorithm, which leverages a quantum generative adversarial network combined with a novel approach for detecting and evaluating abnormal user behavior. Through a feature matching strategy, the algorithm ensures close data alignment between the generator and discriminator. At the same time, the integration of a classical convolutional neural network within the BDE network assesses user behavior abnormalities. Setting distinct thresholds for abnormal behavior and threats enables the differentiation between normal and abnormal activities. Utilizing a dataset and financial stock log data for simulation, the proposed method achieves an AUC value of approximately 0.912 with small negative data samples. Additionally, it records generator and discriminator loss values within the ranges of [1.05,1.55] and [0.49,0.61], respectively, and demonstrates over 80% accuracy in detecting financial stock log anomalies. This method’s reliance on comprehensive big data allows for an in-depth analysis of user behavior, facilitating the timely identification and management of abnormalities.

show abstract

Section: Introductionmentioning

confidence: 99%

Abnormal User Behavior Detection for Industry Big Data Analysis

Wang,

2024

Applied Mathematics and Nonlinear Sciences

View full text Add to dashboard Cite

show abstract

“… Hierarchical clustering methods, in which, a measure of dis‐similarity is used to group or divide, iteratively, vertices into communities. The hierarchical algorithms can be divided into divisive methods 12,13 and aggregation methods 14,15 Approximation methods which optimize the value of a quality function, generally, the function is the modularity, 16 such in greedy algorithm, 14 genetic algorithm, 17, 18 particle swarm optimization, 19 and simulated annealing 20 …”

Section: Introductionmentioning

confidence: 99%

A complex network community detection algorithm based on random walk and label propagation

Benazi¹,

Lounnas

Mokhtari³

2022

Trans Emerging Tel Tech

View full text Add to dashboard Cite

The community structure is proving to have a very important role in the understanding of complex networks, but discovering them remains a very difficult problem despite the existence of several methods. In this article, we propose a novel algorithm for discovering communities in complex networks based on a modified random walk (RW) and label propagation algorithm (LPA). First, we calculate the similarity between nodes based on the new formula of RW. Then, the labels are propagated by the obtained similarity of the first step using LPA. Finally, the third step will be a new measure to find the optimal partitioning of communities. Experimental results obtained on several real and synthetic networks reveal that our algorithm outperforms existing methods in finding communities.

show abstract

“…Numerous studies have attempted to align user accounts from different social networks [1]- [3]. On the basis of overlapping users, many works have also been conducted on multiple social networks, such as item recommendation [4]- [6], influence maximization [7], [8], community detection [9]- [11], and so on. By designing data fusion strategies, scholars are able to depict users more deeply and uncover the differences in user characteristics.…”

Section: Introductionmentioning

confidence: 99%

Understanding User Topic Preferences across Multiple Social Networks

Zhu

Cao

Zhou

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

In recent years, social networks have shown diversity and difference, people begin to use multiple online social networks with the demand for different information content at the same time. The different social networks provide people with kinds of services, and that understanding users' topic preferences across multiple social networks is key to community detection, recommendation, and personalized service across social networks. This paper first divides user topics into two types: global topics and local topics. Global topics are abstracted based on the users' multiple social network data to reflect users' high-level common preferences; Local topics are based on user data from single social network, reflecting users' personalized specific preference influenced by different social networks. On this basis, this paper integrates user behavior data under different social networks, and proposes a user topic preference model MSNT (Multiple Social Networks Topic model) for multiple social networks. The model simulates the interaction process of users across multiple social networks, and outputs users' global topic preferences and local topic preferences synchronously. The model parameters are solved by Gibbs sampling algorithm. This paper uses perplexity, likelihood and PMI-score to verify model performance compared with existing works on data based on well-known sites including Twitter, Instagram and Tumblr.

show abstract

Community detection across multiple social networks based on overlapping users

Cited by 10 publications

References 20 publications

Abnormal User Behavior Detection for Industry Big Data Analysis

Abnormal User Behavior Detection for Industry Big Data Analysis

A complex network community detection algorithm based on random walk and label propagation

Understanding User Topic Preferences across Multiple Social Networks

Contact Info

Product

Resources

About