Graph Convolution Relation Extraction Based on GRU and Attention Mechanism
DU Yan
SUN Yi
Abstract：Entity relation extraction is very important in natural language processing.Aiming at the problems of inaccurate fea-ture extraction in graph convolution network and fuzzy gradient of cyclic neural network,a graph convolution relation extraction mod-el integrating gated cyclic unit(GRU)and attention mechanism is proposed.By adding two-way GRU to process the input context information,more detailed features can be obtained,so as to learn the long-term dependent information,and the multi head atten-tion mechanism is further used to distribute the weight of different types of edges and nodes,filter the redundant information and en-hance the correlation between nodes.Finally,graph convolution is used to get the final relationship extraction result.Experiments on SemEval-2010Task 8 and SemEval-2010Task 4 data sets show that this method improves its F1 value and can effectively extract relationships.
Keywords：relationship extractiongating loop unitattention mechanismgraph convolutional network
Publication Date：2023-11-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 2568-2572,2601 )