From Long News to Accurate Forecast: Importance-Aware Fusion and PRM-Guided Reflection for Time Series Forecasting
A new framework has been developed to enhance time series forecasting by integrating news articles. This approach addresses limitations in existing models by using importance-aware news compression and process-level retrieval supervision. Experiments demonstrate improved prediction accuracy and reduced refinement iterations compared to traditional methods.
- ▪The framework combines importance-aware news compression with process-level retrieval supervision.
- ▪An importance reward model estimates the forecasting utility of news articles for better content preservation.
- ▪A process reward model ranks supplementary news candidates to improve selection quality.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2606.03097 (cs) [Submitted on 2 Jun 2026] Title:From Long News to Accurate Forecast: Importance-Aware Fusion and PRM-Guided Reflection for Time Series Forecasting Authors:Mingyang Liu, Qingcan Kang, Yuke Wang, Shixiong Kai, Kaichao Liang, Hui-Ling Zhen, Tao Zhong, Mingxuan Yuan, Linqi Song View a PDF of the paper titled From Long News to Accurate Forecast: Importance-Aware Fusion and PRM-Guided Reflection for Time Series Forecasting, by Mingyang Liu and 8 other authors View PDF HTML (experimental) Abstract:Incorporating news into time series forecasting is appealing because news can reveal abrupt exogenous events that historical values alone cannot recover.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
