A Semantic Parsing Framework for End-to-End Time Normalization

TL;DR

This paper presents a novel time normalization framework that reformulates the task as code generation within the SCATE semantic framework, enabling improved accuracy, interpretability, and handling of complex temporal expressions.

Contribution

It introduces a code generation approach for time normalization based on SCATE, along with an LLM-based data augmentation pipeline for training effective models.

Findings

Small models trained on augmented data outperform larger LLMs.

The approach handles complex, compositional, and event-relative time expressions.

The framework enables practical, accurate, and interpretable time normalization.

Abstract

Time normalization is the task of converting natural language temporal expressions into machine-readable representations. It underpins many downstream applications in information retrieval, question answering, and clinical decision-making. Traditional systems based on the ISO-TimeML schema limit expressivity and struggle with complex constructs such as compositional, event-relative, and multi-span time expressions. In this work, we introduce a novel formulation of time normalization as a code generation task grounded in the SCATE framework, which defines temporal semantics through symbolic and compositional operators. We implement a fully executable SCATE Python library and demonstrate that large language models (LLMs) can generate executable SCATE code. Leveraging this capability, we develop an automatic data augmentation pipeline using LLMs to synthesize large-scale annotated data with code-level validation. Our experiments show that small, locally deployable models trained on this augmented data can achieve strong performance, outperforming even their LLM parents and enabling practical, accurate, and interpretable time normalization.