Information Security Field Event Detection Technology Based on SAtt-LSTM

Detecting information security events from multimodal data can help analyze the evolution of in field. The Tree-LSTM network that introduces self-attention mechanism was used to construct sentence-vectorized representation model (SAtt-LSTM: with self-attention) and then classify candidate event sentences through results SAtt-LSTM obtain sentence types. Event detection using classification methods solve problem error cascade based on pipeline methods, CNN or RNN cannot make full use syntactic joint learning. paper treats task as a task. In order verify effectiveness superiority method this paper, DuEE set for experimental verification. Experimental show has better performance than chain structure LSTM, CNN, only Tree-LSTM.