Consumer Sentiment Analysis and Product Improvement Strategy Based on Improved GCN Model

In recent years, the amount of online shopping review data has increased dramatically. Obtaining information that helps business decision-making from such complex and massive reviews has become a difficult and important task for merchants. This paper uses sentiment analysis technology to innovatively introduce the attention mechanism on the LSTM infrastructure of the baseline model, and proposes a word vector structure and a BiGRU structure to build an online user sentiment analysis system based on deep learning. The system includes a user review sentiment classification and sentiment analysis model based on the improved GCN model. The experimental results also show the superiority of our method, which brings 4.73%, 7.84% and 5.72% F1-score improvements to the algorithm respectively. It proves that the two algorithms proposed in this paper can effectively achieve their goals and achieve high performance.