Data Generation Method and Training Mode of Relationship Extraction Based on Neural Network

Abstract

As one of the most common and effective techniques for obtaining information, relation extraction is a key task in machine learning. Understanding the current development status of computer technology, we know that the technical methods of extracting information mainly rely on a large number of manual operation and processing features, and with the full promotion of neural networks, it provides a new perspective for actual relationship extraction and data generation. Therefore, on the basis of understanding the current relationship extraction and data generation methods, this paper conducts an empirical analysis of the Chinese character relationship extraction method with convolutional neural network as the core and designs and proposes a Chinese character relationship extraction system scheme based on the convolutional neural network model. Using the convolutional neural network model to automatically obtain features and classify the relationship between the characters mastered, the actual accuracy rate can reach 92.87%, and the average recall rate can reach 86.92%. The final result proves that the relationship extraction data generation method based on neural network has unique application advantages.