Overview
ISO 24617-11:2021 - "Language resource management - Semantic annotation framework (SemAF) - Part 11: Measurable quantitative information (MQI)" defines a standardized way to represent measurable quantitative information (MQI) in text and language resources. The standard specifies a Quantitative Markup Language (QML) model (abstract and concrete syntaxes) to capture measures as pairs , normalize units, and make annotations interoperable across NLP pipelines. It targets practical measurement data in scientific, technical, and applicational domains for use in IR, QA, text summarization and other NLP tasks.
Key topics and technical requirements
- QML Metamodel and Abstract Syntax (QML_as): an abstract specification for MQI annotations that supports normalization and interoperability across serialization formats.
- Concrete syntaxes:
- XML-based concrete syntax (QML_csx) - standoff XML serialization (includes tag names with ID prefixes, root attributes, element and link attribute specifications).
- TEI-based concrete syntax (QML_cst) - TEI-compatible serialization for corpora using TEI conventions.
- Annotation primitives and links: explicit element types and link types to connect numeric values, units, measured entities, contexts (e.g., ranges, approximations).
- Unit handling and normalization: treatment of base and derived units (SI and conventionally used units), and guidance for representing units consistently (annex on unit representation).
- Interoperability: mapping and compatibility with other SemAF parts for temporal durations and spatial measures, and alignment with LAF (ISO 24612) and lexical/morpho-syntactic frameworks.
- Standoff annotation: recommendations to keep annotations separate from primary data for flexible downstream processing.
- Informative annexes with sample data, annotation procedures, and informal MQI statements to guide implementers.
Practical applications
ISO 24617-11 is designed for applied use in systems that must extract, normalize, compute or aggregate measurements:
- Information retrieval (IR): indexing and searching documents by numerical attributes (e.g., product specs, KPIs).
- Question answering (QA): precise extraction of numeric answers (e.g., “What is the dosage?”).
- Text summarization (TS): preserving and normalizing key measures in summaries (e.g., financial reports).
- Medical informatics & life sciences: annotating doses, lab values, eligibility criteria in clinical texts.
- Business analytics: aggregating financial figures across reports with normalized units.
Who should use it
- NLP engineers, computational linguists, data scientists, corpus linguists
- Developers of IR, QA and summarization systems
- Medical informatics and BI teams processing technical documents
- Standards teams and tool builders needing interoperable measurement annotation
Related standards
Keywords: ISO 24617-11, MQI, QML, semantic annotation, measurable quantitative information, language resource management, SemAF, NLP, information retrieval, question answering, text summarization, units, XML, TEI.