Dependency information integrated news event time sequence relation identification method

• Main Authors: ZHANG YAFEI, GUO JUNJUN, LI LIANGYI, YU ZHENGTAO, GAO SHENGXIANG
• Format: Patent
• Language: Chinese; English
• Published: 30.08.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention relates to a news event sequential relationship identification method integrated with dependency information, and belongs to the technical field of natural language processing. According to the method, an event sentence is converted into a vector sequence through a pre-training model, the sentence vector sequence is sent into a Bi-LSTM, context information is fused into the sentence vector sequence, event semantic information is extracted from the sequence, then dependency syntax information of the event sentence is extracted through two-way dependency attention, and the dependency syntax information of the event sentence is extracted through two-way dependency attention. Using event semantic information and dependency information to jointly represent time sequence relation characteristics of the events, and finally judging the occurrence sequence of the events through a multi-layer perceptron; according to the method, event sequential relation identification is carried out based on Bi-LSTM and