Process / pipelineText miningPipeline

Temporal Expression Extraction (TIMEX)

Also known as: TIMEX, temporal tagging, TIMEX3 extraction, Zamansal İfade Çıkarma (TIMEX)

Sources2Related methods2

Temporal expression extraction is a natural-language-processing task that detects dates, times, durations, and frequencies in text and normalises them to the TimeML/TIMEX3 standard. Building on the TempEval shared task introduced by Verhagen et al. (2007), it turns time references scattered through free text into structured, machine-readable values that support event timelines and chronological analysis.

Key highlights

  • Normalises diverse, free-form time references to a single TIMEX3 standard, making them directly comparable.
  • Resolves relative expressions ('yesterday', 'next month') against a reference date into concrete values.
  • Feeds directly into event timelines and chronological analysis across documents.

Intuition

This section is available to Pro members. Upgrade to Pro

How it works

This section is available to Pro members. Upgrade to Pro

When to use it

Use temporal expression extraction when you have text and want to recover the dates, durations, and frequencies it mentions in a normalised, comparable form — for example to build event timelines or run chronological analysis. The text should be in English or another supported language, and the reference (context) date must be known so that relative expressions can be resolved. A modest corpus is enough to start.

Strengths & limitations

Strengths
  • Normalises diverse, free-form time references to a single TIMEX3 standard, making them directly comparable.
  • Resolves relative expressions ('yesterday', 'next month') against a reference date into concrete values.
  • Feeds directly into event timelines and chronological analysis across documents.
Limitations
  • Works only for English or other explicitly supported languages.
  • Requires a known reference (context) date; without it, relative expressions cannot be normalised.
  • Implicit or vague time references that lack a clear anchor remain hard to resolve.

Common pitfalls

This section is available to Pro members. Upgrade to Pro

Applications

This section is available to Pro members. Upgrade to Pro

Frequently asked

What is TIMEX3?

TIMEX3 is the temporal expression markup of the TimeML standard. It defines how dates, times, durations, and frequencies found in text are encoded as normalised, machine-readable values, so that 'next Friday' becomes a concrete calendar value rather than raw text.

Why does it need a reference date?

Many expressions are relative — 'yesterday', 'two weeks ago', 'next month'. They can only be resolved into an explicit value relative to a known context (reference) time. Without that anchor, relative expressions cannot be normalised.

Does it work in languages other than English?

It works for English and other explicitly supported languages. Multilingual and cross-domain temporal tagging (Strötgen & Gertz, 2013) broadened coverage, but applying it to an unsupported language degrades recognition.

What is the difference between detection and normalisation?

Detection finds the spans of text that mention a time, while normalisation maps each detected span to a standardised TIMEX3 value. Detection alone tells you where a time is mentioned; normalisation makes that mention comparable across the document.

Sources

  1. 1.
    Verhagen, M. et al. (2007). SemEval-2007 Task 15: TempEval Temporal Relation Identification.
  2. 2.
    Strötgen, J. & Gertz, M. (2013). Multilingual and Cross-Domain Temporal Tagging. Language Resources and Evaluation (LRE).

You have read it. What now?

Cite this page

ScholarGate. (2026, June 1). Temporal Expression Extraction. ScholarGate. https://scholargate.app/text-mining/temporal-expression-extraction