Sensitive Word Recognition Scheme Based on Text-RCNN Model

The ocean is a gateway between the countries of the world and a treasure trove of resources for the future life of society. The development of new technologies has promoted the development of the ocean, but it has brought about many problems in the time. To further govern ocean security, it is necessary to use a new generation of scientific and technological means to solve these problems. In this paper, we will design a Text-RCNN based text-phonemes dual-channel sensitive word recognition scheme. The solution processes speech in the preprocessing stage to make it easy to recognize words and phonemes, then builds a text Text-RCNN model and a phonemes Text-RCNN model, and fuses the two models in the SoftMax laye. Comparing this paper's scheme with the traditional sensitive word recognition model, the experimental results show that the accuracy and precision of this paper's scheme are improved.