Temporal feature-enhanced hierarchical news context generation and visualization system using F-IWF and K-means

To address the issues of discrete event elements, ambiguous evolutionary logic, and obscured deep connections in multi-source news data, this paper proposes a hierarchical news context generation and intelligent visualization method based on progressive feature processing and temporal modeling. This method employs TF-IWF for feature weighting, combines TextRank to select core keywords, uses K-means for topic clustering, and utilizes Bi-LSTM to extract temporal features to generate a tree-like temporal context. Finally, a visualization system is built based on ECharts. Experimental results show that the method achieves an accuracy, recall, and F1 score of 95.89%, 92.97%, and 94.05%, respectively, with a temporal consistency accuracy of 93.67%. The readability and completeness scores for hot topics are 4.35 and 4.89, respectively, and the average system response time under high concurrency is 279.49 ms. This method integrates discrete news into a clear temporal sequence and hierarchical tree structure, providing an efficient tool for news analysis and public opinion assessment.