Assisted Analysis of Big Data Algorithms for Public Security in Crime Governance

With the high-quality development of society, public safety is an important control direction, and the use of computers to assist in intelligent crime warning is an important way. This study combines the multi-modal information of text data streams, video data and audio data from the system level to build an intelligent crime warning model algorithm. The proposed algorithm can improve the accuracy of the crime monitoring and enhance the effect of smart management and control. In the text data stream intelligent analysis module, this study designed an abnormal data stream detection algorithm based on function signature matching. This algorithm can effectively identify potential security threats in the network through complex data streams. In the video data module, this study proposes an improved YOLOv11 target detection algorithm. The algorithm uses GIOU Loss to optimize the target box regression loss, which enhances the detection ability of small targets to a certain extent, thereby realizing real-time warning and analysis of abnormal human behavior. In the audio data analysis module, this study combines den-oising algorithms and spectrum analysis techniques to improve the accuracy of audio abnormal signal detection. The designed experiment was conducted on a real public safety data-set and analyzed through separate tests of data stream, video, and audio. The results showed that the proposed method has a stable warning accuracy under multi-modal information fusion.