ResearchResearch paperLarge Language Models · Training & Scaling1 source · Oct 10, 2026

World2TS: Real-World Event Context for Time-Series Forecasting

We present World2TS, a scalable construction framework and large-scale dataset for time-series forecasting with real-world event context.

Time-series forecasting plays a vital role in a wide range of real-world applications.
In practice, time series are observations of complex real-world processes, and their trajectories may reflect events, interventions, disruptions, and changes in measurement or reporting that cannot always be inferred from past values alone.
A central barrier to studying context-aware forecasting is therefore the lack of large-scale datasets that align real-world events with the individual series they affect.
World2TS provides a reusable data foundation for developing and evaluating context-aware forecasting models and agents.

[1]World2TS: Real-World Event Context for Time-Series Forecasting
arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 10, 12:31 PM
We present World2TS, a scalable construction framework and large-scale dataset for time-series forecasting with real-world event context.
Time-series forecasting plays a vital role in a wide range of real-world applications.

Extractive summary: sentences quoted from the sources.

Before this