NUM2EVENT: Interpretable Event Reasoning from Numerical time-series

TL;DR

This paper introduces NUM2EVENT, a framework that enables interpretable event reasoning directly from numerical time-series data, addressing the gap in LLM understanding of purely numerical signals.

Contribution

The work presents a novel task of number-to-event reasoning, along with a reasoning-aware framework that combines event extraction, synthetic data generation, and a two-stage fine-tuning pipeline.

Findings

Outperforms strong LLM baselines in event-level precision and recall

Explicitly reasons over numerical changes and generates explanations

Bridges quantitative reasoning with semantic understanding

Abstract

Large language models (LLMs) have recently demonstrated impressive multimodal reasoning capabilities, yet their understanding of purely numerical time-series signals remains limited. Existing approaches mainly focus on forecasting or trend description, without uncovering the latent events that drive numerical changes or explaining the reasoning process behind them. In this work, we introduce the task of number-to-event reasoning and decoding, which aims to infer interpretable structured events from numerical inputs, even when current text is unavailable. To address the data scarcity and semantic alignment challenges, we propose a reasoning-aware framework that integrates an agent-guided event extractor (AGE), a marked multivariate Hawkes-based synthetic generator (EveDTS), and a two-stage fine-tuning pipeline combining a time-series encoder with a structured decoder. Our model explicitly reasons over numerical changes, generates intermediate explanations, and outputs structured event hypotheses. Experiments on multi-domain datasets show that our method substantially outperforms strong LLM baselines in event-level precision and recall. These results suggest a new direction for bridging quantitative reasoning and semantic understanding, enabling LLMs to explain and predict events directly from numerical dynamics.