A Graph Convolutional Network-Based Sensitive Information Detection Algorithm

Algorithm 1

The Proposed GCSA algorithm.

Input. Original document; sensitive word corpus;
	(1) Preprocess the original text;
	(2) Construct the sensitive word corpus into a directed graph of sensitive words;
	(3) Obtain the word embedding;
	(4) Apply the graph convolutional network and self-attention mechanism to extract structural information and relationship from keywords;
	(5) Predict the probability of candidate keywords;
	Output. The probability of candidate keywords;