Research on Clue Mining in Criminal Cases of Smart Phone Trojan Horse under the Background of Information Security. (25th February 2022)

• Record Type: Journal Article
• Authors: Gang, Li; Wen, Yong
• Other Names: Zhong Shan Academic Editor.
• Is Part Of: Journal of robotics. Volume 2022(2022)
• Publication Date: 2022-02-25
• Subjects: Robotics -- Periodicals; Robotics; Periodicals; 629.892
• DOI: 10.1155/2022/7568110
• Languages: English
• ISSNs: 1687-9600

Abstract

Abstract : With the massive popularity and wide application of Android smartphones, there are more and more malware targeting Android smartphones. Research and analysis Android smart phone Trojan horses can provide corresponding technical support for malware detection on Android smart phones, which has good scientific research significance and broad market value. This work studies and analyzes the existing implantation technology of Android smart phone Trojans and analyzes the basic principles and implementation methods of obtaining root permissions on mobile phones. At the same time, the basic principles and implementation methods of mobile phone Trojan horse hiding are also studied. Through the research of the broadcast receiver model of the Android platform, the background monitoring principle and implementation technology of the mobile phone Trojan horse are analyzed, and the theoretical foundation and technical support are provided for the implementation of the Trojan horse background monitoring program in this article. Aiming at the problem of insufficient training corpus in the event relationship classification task, this work proposes an event relationship classification method based on tritraining. This method first trains three different classifiers based on the labeled dataset. In the collaborative training process, the new labeled event pairs used to expand each classifier are provided by the other two classifiers. For the same unlabeled event pair, the relationship prediction results are consistent; then, the event pair is considered to have a higher classification confidence and is placed in the labeled set of the third classifier after labeling. Finally, a well-trained classifier is used to determine the relationship between the pair of events to be tested by voting. This study constructs a weighted network structure model called the conceptual network and determines its upper weight based on the structural information and text data of the knowledge network. Aiming at the problem of the lack of means for mining-related forms between things, the pheromone strategy of absorbing the ant colony algorithm is proposed, and random walks are performed on the conceptual network. By analyzing the pheromone distribution information in the convergent state, the calculation of the semantic relevance is completed. At the same time, the method of semantic clue discovery is realized. The experimental results show that the human cognitive information contained in the knowledge network can meet the needs of the mining of the related forms of things, and the performance of the semantic correlation calculation method based on the convergence pheromone is close to other random walk methods based on the knowledge network.