A Novel Model Based on Big Data Environment for Text Content Security Recognition

In the big data environment, text content security recognition is one of the main ways to intelligently manage the Internet and maintain privacy. However, traditional text content security recognition methods lack semantic understanding and ignore scenarios where keywords are evenly distributed, resulting in high false positive rate and low accuracy. To address this problem, we propose a novel model based on big data environment for text content security recognition. In the scenario where keywords are evenly distributed, we design the TFC-BPLW-AM algorithm to extract text vectors. The TFC BPLW-AM algorithm considers the problem of uniform distribution of keywords, the problem of calculating weights in a single form, and the time-consuming problem caused by too large weight matrix. Thus, the weight integrity is enhanced, the recognition accuracy is improved, and the running time is shortened. Under the 20 newgroups and Fudan University Chinese text datasets, we conduct experimental comparisons with existing models and results show that our model achieves 96.7% F1 score, with a maximum increase of 30.7% and a minimum increase of 2.7%.