Description
With the rapid development of social media, Weibo and other platforms have become important platforms for information propagation. This research focuses on the problem of sentiment classification on Weibo platform, aiming to analyze the influence of different kinds of sentiment (such as positive, negative or neutral) on information propagation, and to build an efficient prediction model that predicts the sentiment classification in users’ comment content. Given the complexity and diversity of social network data, traditional machine learning methods face many challenges in addressing such issues. Therefore, a multi-model fusion which combines Graph Neural Network (GNN) and Long-Short Term Memory (LSTM) is proposed in this paper. Graph Neural Network (GNN) is good at capturing spatial dimension information, while Long-Short Term Memory (LSTM) is suitable for time series analysis. The combination of these two models can significantly improve the accuracy of sentiment classification prediction. This research has found that the information propagation mechanism of Weibo has a strong tendency of community structure, and positive information is more likely to be widely forwarded and commented in the community, which provides a basis for information propagation mechanism based on community structure. In addition, the results show that the multi-model fusion strategy significantly improves the prediction performance in the classification issues. What’s more, this research has also discussed the influence of model hyperparameter adjustment on sentiment classification, analyzed the optimization effect of different parameter settings, so as to find the best parameter combination of each model, thereby providing more scientific support for sentiment classification models. This research has identified a new multi-model fusion method for Weibo platform, which captures the features of information from spatial and temporal multi-dimensions, and greatly improves the performance of a single model.| Period | 1 Jan 2024 → 24 Dec 2024 |
|---|---|
| Degree of Recognition | International |