Abnormal event-oriented relationship extraction method

Abstract

The invention discloses an abnormal event-oriented relationship extraction method, which comprises the following steps of: arranging an entity relationship data set related to an emergency according to a structured triad form, and converting statements in related fields into vectorized expressions; constructing a shared coding layer by using a bidirectional long-short-term memory network in combination with a self-attention mechanism; predicting the annotation of the main entity by using a softmax function, performing shared coding on the relational customer entity by using a convolutional neural network, and enhancing coding representation through a prediction result of the main entity; and optimizing the training parameters by using the self-attention mechanism again. A conflict problemof multiple pairs of entities and relationship categories in the emergency text can be well solved, and the extraction quality of the text entity relationship in the field can be improved.

Description

technical field [0001] The invention relates to the field of relation extraction in natural language processing technology, in particular to a relation extraction method oriented to abnormal events. Background technique [0002] With the continuous development of the global economy and the continuous increase of the world population, the number of tourists in the scenic spot is increasing, so the scenic spot has become a highly populated place, making it prone to various abnormal events that affect the order of the scenic spot and even cause serious damage to public property. loss. At present, the research on abnormal events in scenic spots mainly uses monitoring methods to prevent them. However, various abnormal events are complex and changeable. Real-time monitoring can only make corresponding preparations after the event occurs. Provide decision-making assistance for abnormal events in the subsequent processing process. Entity relationship extraction is one of the core ...

AI Technical Summary

Problems solved by technology

[0004] The current mainstream information extraction models cannot cover the important features of information extraction well.

If there are more than two entities appearing at the same time, the system model based purely on entity recognition has the problem of low recall rate and poor sampling rate when facing one-to-many samples; The influence of entity overlap cannot be solved for many samples, and the extraction efficiency is also greatly reduced

Images