Relation Extraction of Minority Cultural Information Resources based on Bi-LSTM and Double Attention Mechanism

In recent years, the research of minority cultural information resources based on knowledge graph has become a research hotspot in academia. Relation extraction is an integral part of building a knowledge graph. In order to improve the accuracy of the relationship extraction of minority cultural information resources, this article first introduces Bidirectional Long Short-Term Memory (Bi-LSTM) to extract feature values. Secondly, it adds word-level and sentence-level attention mechanism to reduce the noise interference of the text. Finally, on the minority cultural information resources contrast experiments were carried out. The experimental results show that the Bi-LSTM neural network relation extraction model with double attention mechanism has better performance in accuracy, recall rate and comprehensive value index (F-Measure).