期刊名称：IAENG International Journal of Computer Science
印刷版ISSN：1819-656X
电子版ISSN：1819-9224
出版年度：2019
卷号：46
期号：4
页码：616-627
出版社：IAENG - International Association of Engineers
摘要：Current Online Social Network (OSN) needs realtimeand adaptive security model. The tremendous success ofdeep learning algorithms at computer vision tasks in recentyears inspires us to adopt the method. It is becomingincreasingly popular for various applications include in OSNsecurity and privacy-preserving. In this paper, we proposeDeepSentiment, a dynamic deep learning model to detect andclassify malicious sentiment in OSN. Different fromconventional CNN, we introduce RunPool, a dynamic poolingfunction to train the sentiment features. By using the function,we find a significant increase in the graph’s performance withthe DeepSentiment CNN model. Demonstrated by theexperiment, we harvest a higher accuracy and small loss inmalicious sentiment classification with the benchmark dataset.
关键词：Neural Networks; Malicious Sentiment;Dynamic Deep Learning; Online Social Network
