Research on Classification Algorithm of Graph Neural Network Based on Combination of Graph Neural Network and LSTM

This paper mainly discusses the application of graph neural network combined with long short-term memory (LSTM) model in text classification. The aim of this study is to integrate the advantages of graph neural networks in processing complex dependencies between non-Euclidean data and capture nodes, and the superior ability of LSTM in network sequence modeling and long-term dependency capture. We propose an innovative framework that uses graph neural networks to extract and encode node features in graph structures, and then uses LSTM to learn the attribute information of these high-dimensional and structured feature sequences and construct graph data to realize efficient classification of complex graph data. In the experiment, we apply this algorithm to a number of actual scene tasks with rich node relationships, and compare it with the existing mainstream classification methods. The results show that this classification algorithm, which combines graph neural network and LSTM, can better understand and utilize the topological structure information in graph data while maintaining high classification accuracy, thus improving the overall performance and generalization ability, and providing a new and effective way for in-depth mining and understanding of complex graph data.