Article Preview
Top1. Introduction
At present, there are few mobile application recommendation systems based on user trust behavior in existing work (Ramlatchan, 2018). The trust behavior of users using mobile applications can truly reflect their personal preferences, and their data is of great value for constructing user profiles and calculating recommendations. Secondly, the accuracy and personalization of existing recommendation systems still need to be improved. Existing mobile stores are basically scoring rules based on the number of downloads of applications and user evaluation, and such rules have the risk of low accuracy and malicious attacks. Last but not least, the issue of privacy protection in the mobile application recommendation system is still an open issue (Ghezzi et al., 2015). Since the recommendation result is based on the calculation of user data and even contains a lot of privacy data, if the user data cannot be properly protected, privacy leakage will occur.
The current Android system widely uses the permission model technology to manage and control the acquisition and access to information related to user privacy information (Li et al., 2019). However, the widespread application of mobile application platforms also has the problem of security and abuse of permission models. Many mobile applications often apply for unnecessary sensitive information permissions, so that users ’private information-related information often faces a huge risk of being leaked by malicious intentions. At the same time, many mobile applications will actively obtain and maliciously disclose the user's private information related information without the knowledge of the developer and the user. In recent years, a lot of research and work have begun to focus on the analysis of applications and the protection of the security and privacy of mobile application systems. Whether the permission of sensitive information behavior of the application is maliciously disclosed and whether it should be clearly allowed by the relevant departments, so a reasonable privacy information protection scheme using sensitive information permissions should be set up.
In response to the question of whether mobile phone privacy information may be exposed, M et al. Modified the software and used a virtual machine to implement dynamic mobile phone application flaw data analysis tool Taint-droid (Kumar et al., 2016; Xu et al., 2020). The advantage of this data analysis tool is that it can directly mark sensitive data as a taint source, and then track the data marked as taint at runtime, and analyze and judge based on the data marked as taint source and whether the data is leaked by the application Whether there may be leakage of mobile phone privacy information in the process of running the application (Wang et al., 2016). The study provides a continuous and automated application risk assessment framework, which automatically builds application models from mobile phone application metadata by automatically collecting mobile phone user responses to application permissions and use, and then uses a machine deep learning Methods to analyze and assess the risk of application (Wan et al., 2019). The research of Wei et al. Proposed that crowdsourcing data analysis technology can be used to study each user's expectation and acceptance of different privacy information combinations (Wei & Hou, 2016). By using this technology to detect the central expectations of each application and the amount of privacy information as well as the use rights and location, you can directly analyze the application core privacy information expectations and solutions for the use of information and behavior, That is used in third-party databases, or is required for the core functions of the application software (Ding et al., 2016). However, the results of the research indicate that the core expectations of users and the expectations and plans for the use of private information also directly affect each user's expectations and acceptance of the application's core private information and behavior.