Modeling and Simulation of User Survival Space in UGC Platforms Integrating Sentiment Analysis and Behavior Prediction
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
In response to the problem that the 30-day retention rate of ordinary users on UGC platforms (such as TikTok, Weibo) is less than 40%, and the existing user modeling lacks the integration of emotional dimension and behavior prediction, this paper constructs a user survival space model that integrates "emotion - behavior - social - content" four dimensions to achieve quantitative assessment of user status and retention prediction. The study uses 1000 ordinary Weibo users (data from October to December 2023) and 1200 ordinary TikTok users (data from September to November 2023) as samples. Through the BERT model, sentiment analysis is completed (the accuracy rate of Weibo data is 88.6%, and that of TikTok data is 86.3%), combined with the LSTM time series model to predict the future 30-day user behavior intensity (the MAE of Weibo data is 0.72, and that of TikTok data is 0.68), and using the Analytic Hierarchy Process (AHP) to determine the weights of each dimension (the behavior dimension is 0.30, the emotional/social dimension is each 0.25, and the content dimension is 0.20). Finally, the user survival space index (U ∈ [0, 100]) is formed. Simulation experiments show that the survival space index of this model has a Pearson correlation coefficient of 0.75 (TikTok) - 0.78 (Weibo) with the actual retention days of users, and the retention prediction F1-score is 0.80 (TikTok) - 0.82 (Weibo), which is 7-8 percentage points higher than the baseline model without an emotional dimension or behavior prediction. This model can provide a quantitative tool for UGC platform user churn warning and personalized operation, and enriches the theoretical application of user behaviorology and computational sociology in platform ecosystem research.
Fields of Study & Subjects
Customer churn and segmentation
Online Learning and Analytics
AI and HR Technologies
Publication details
- Journal / source
- IEEE International Conference on Power, Electronics and Green Energy (ICPEGE)
- Pages
- 1144–1148
- Publication date
- 28 January 2026
- Language
- EN