Neural Network Based Target User Recognition Model for Network Community

6th Annual 2018 International Conference on Geo-Spatial Knowledge and Intelligence

Wu, Xu^1^2^3 ; Dai, Yulun^1^2^3 ; Xie, Xiaqing^1^2^3

Key Laboratory of Trustworthy Distributed Computing and Service, Ministry of Education, China^1

School of Cyberspace Security, BUPT, China^2

Beijing University of Posts and Telecommunications Library, Beijing

100876, China^3

关键词: Classification algorithm;    Classification methods;    Construction costs;    Improve performance;    Network communities;    Recognition models;    User behavior analysis;    User classification;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/234/1/012075/pdf
DOI  :  10.1088/1755-1315/234/1/012075

来源: IOP

【 摘 要 】

For multi-dimensional feature of user data in network community, available methods mainly use Rank scoring algorithm or user classification algorithm for target user recognition. However, Rank method has a low performance, and the classification algorithm needs high construction cost. Therefore, this paper uses a target user recognition model integrating the user content analysis and the user behavior analysis to improve performance and speed of the target user recognition by a neural network content analysis model with a single-layer neural network and N-gram features for discovering automatically the user feature. The proposed method outperforms the current Rank scoring and classification methods in terms of performance, in which the F value reaches 0.89 and the accuracy reaches 0.91. Moreover, avoiding the cost of manual design dependent on specific tasks shortens the training time. Ten thousand data can be modeled in one minute.