BERT-Embedding-Based JSP Webshell Detection on Bytecode Level Using XGBoost

Algorithm 1

Word segmentation algorithm of BERT.

	Input: A word to be split, W
	Output: Lists of subwords, L
	while do
	.
	if L is empty then
	L.push(S)
	else
	L.push(strcat(“##,”S))
	end
	.
	end