Contextual Analysis Using Deep Learning for Sensitive Information Detection
Yaxin Liang, Erdi Gao, Yuhan Ma, Qishi Zhan, Dan Sun, Xingxin Gu · 2024
This project intends to study A semantic feature extraction algorithm that combines semantic feature-oriented lexical vector representation with attention mechanism (A-ELMO). Firstly, the fast comparison of the dictionary tree reduces the comparison of invalid features as much as possible and greatly improves the speed of retrieval. A language pattern lexical vector (ELMo) model for context analysis is established. The method uses a dynamic word vector to describe the context, which improves the extensibility of the system. The attention mechanism is introduced to enhance the recognition degree of the sensitive features in the image. We improved detection rate of sensitive data. Compared to the traditional word-level emotion parsing algorithm, the proposed algorithm shows significant improvements in accuracy. Additionally, the accuracy of this algorithm surpasses that of traditional keyword comparison algorithms. The experimental results demonstrate that the proposed algorithm has clear advantages in enhancing the sensitivity and detection ability of the sensor.