Overview
ISO 24617-4:2014 - Language resource management - Semantic annotation framework (SemAF) - Part 4: Semantic roles (SemAF‑SR) defines a consensual annotation scheme for semantic roles. It provides a metamodel, terminology and guidance for marking the role a participant plays in an eventuality (event, state, action) as expressed mainly by verbs and other predicative elements (nominalizations, nouns, adjectives, predicate modifiers). The standard supports interoperable semantic role annotation (SR / SRL) across corpora and tools while explicitly excluding the predicating role of adverbs and the use of coercion.
Key technical topics and requirements
- Core terminology and formal units: precise definitions for arguments, predicates, predicate-argument structures, entities, participants, eventualities and semantic roles.
- Metamodel for semantic role annotation: abstract/concrete specifications for representing roles and their relations (see Clause 6 and Annex C for annotation language syntax).
- Eventuality frames: generalized frame-based specifications linking predicates to expected arguments and semantic roles.
- Criteria and guidelines: methods for distinguishing semantic roles, handling fuzzy boundaries, inheritance and multiple classification (Clauses 6–7).
- Interoperability & mappings: guidance for mapping existing frameworks (FrameNet, VerbNet, PropBank, LIRICS, EngVallex) to the ISO metamodel and for developing new role frameworks for languages or domains (Clause 8, Annex B).
- Normative and informative material: Annex A lists the ISO semantic roles specification; Annex B reviews existing frameworks; Annex C describes the annotation language.
Practical applications
ISO 24617-4:2014 is directly applicable to:
- Corpus annotation and building gold-standard datasets for semantic role labelling (SRL).
- NLP pipelines for information extraction, question answering, event extraction, semantic parsing and machine translation where consistent participant-role semantics are needed.
- Lexicon and frame development (linking word senses to eventuality frames).
- Tool interoperability and resource sharing, enabling mappings across FrameNet, PropBank and VerbNet-style resources.
- Research in computational linguistics, semantic web, and dialogue systems that require standardized role semantics.
Who should use this standard
- Computational and corpus linguists, language resource developers, annotation project managers, and SRL tool builders.
- Standards bodies, academic researchers and industry teams aiming for consistent, interoperable semantic annotations across languages and domains.
Related standards / ecosystem
- ISO 24617 series (SemAF parts for Time, Dialogue Acts, Discourse Structure, Spatial information)
- FrameNet, VerbNet, PropBank, EngVallex, LIRICS (frameworks referenced for mappings)
Using ISO 24617-4:2014 (SemAF‑SR) improves annotation consistency, enables cross-resource mappings, and supports robust SRL applications across NLP tasks.