Research on Automated Review Method for Harmful Information of Smart Device

As an important information exchange platform, smart device applications are generating a huge amount of multimedia data anytime and anywhere, the harmful information in the data will endanger the physical and mental health of users. In response to this issue, the paper proposes an automated review method for harmful information from smart device applications, which can review and rate the content interacted with smart device applications, thereby grading the applications, identifying the applications that affect the physical and mental health of users, and extracting all detected harmful content as evidence, and then saving it in the abnormality report. For multimodal content, this paper also proposes a new multimodal fusion algorithm for the fusion of multimodal content, which facilitates the rating and review of applications content. The results have shown that by rating and reviewing multimodal fusion content, the harmful information from applications can be well distinguished, thereby protecting the physical and mental health of users.