ISO 24617-6:2016

Overview

ISO 24617-6:2016 - Language resource management - Semantic annotation framework - Part 6: Principles of semantic annotation (SemAF Principles) - defines the methodological and conceptual basis of the ISO Semantic Annotation Framework (SemAF). It documents the strategy for developing separate, interoperable annotation schemes for discrete semantic phenomena (e.g., time/events, dialogue acts, semantic roles) with the long‑term goal of integrating these into a single, coherent wide‑coverage semantic annotation scheme.

Key themes in the standard include the distinction between abstract syntax and concrete syntax for annotations, the specification of a metamodel, and the requirement for a clear semantic interpretation of annotation objects. The standard also addresses consistency across SemAF parts and gives guidance on cross‑cutting semantic phenomena such as negation, modality, and quantification.

Key topics and technical requirements

• Purpose and motivation: promote methodological coherence across SemAF parts and reduce risks arising from separately developed schemes.
• Metamodel: define schematic concepts and relationships used in annotation schemes (conceptual backbone for interoperability).
• Abstract vs. concrete syntax: separate the annotation’s conceptual structure (abstract) from its representational format (concrete), mirroring the ISO Linguistic Annotation Framework (ISO 24612).
• Semantic interpretation: require well‑founded semantics for annotations to ensure consistent meaning across resources.
• Design methodology: staged steps for creating annotation schemes, including feedback loops and optional elements to increase reuse.
• Handling overlaps: procedures to identify and mitigate terminological or conceptual inconsistencies between parts (e.g., spatial/temporal relations appearing as semantic roles).
• Cross‑cutting phenomena: guidance for annotating modalities, polarity/negation, quantification, measures and attribution when these cut across different SemAF parts.

Practical applications and target users

Who benefits:

• NLP and computational semantics researchers building annotated corpora.
• Corpus linguists and language resource developers creating interoperable semantic annotations.
• Annotation-tool and platform developers implementing standards‑compatible formats (abstract/concrete syntax separation helps tool interoperability).
• Machine‑learning engineers training models for event extraction, temporal reasoning, question answering, dialogue systems and semantic parsing.
• Standards bodies and project managers coordinating multi‑part annotation schemes.

Practical advantages:

• Improved interoperability of semantic resources across projects and tools.
• Clearer foundations for combining annotations (e.g., time, spatial info, semantic roles) to support higher‑level applications such as text‑based question answering or automated semantic extraction.
• Reduced ambiguity when multiple SemAF parts interact, making datasets more robust for ML training and evaluation.

Related standards

• ISO 24617 series (SemAF): Part 1 (Time and events, ISO‑TimeML), Part 2 (Dialogue acts), Part 4 (Semantic roles), Part 5 (Discourse structures), Part 7 (Spatial information); Parts 8 and 9 under preparation.
• ISO 24612: Linguistic Annotation Framework - referenced for the annotations vs representations distinction.

For authoritative details and purchasing the full text, consult ISO (www.iso.org) or your national standards body.