In the current processing of semantic relations, traditional deep learning methods have weak context-dependent issues. This paper proposes a relation processing method based on the combination of GRU model and word vector splicing. After vectorizing the text information, the word vector is reconstructed with the context of the word, and separate from the small sample set of data to do targeted feature extraction, classifying by bidirectional GRU model, and finally adopting attention mechanism to further improve model classification accuracy. Based on the actual customer service dataset of a certain city-level power grid for verification and comparison experiments, the results show that the model can effectively improve the accuracy of text semantic classification.
展开▼
