Overview
ISO/IEC 5259-1:2024 - Artificial intelligence - Data quality for analytics and machine learning (ML) - Part 1: Overview, terminology, and examples is the foundational part of the ISO/IEC 5259 series. Published July 2024, this first edition establishes a common conceptual basis and standardized terminology for data quality in the context of analytics, machine learning (ML) and AI projects. It clarifies how the parts of the 5259 series relate, introduces core concepts, and provides illustrative examples, use cases and usage scenarios to support consistent interpretation across organizations.
Key topics and technical coverage
ISO/IEC 5259-1 focuses on the following technical topics:
- Terminology and definitions: standardized terms such as data life cycle, data originator, data holder, data user, data quality, feature, data provenance, and data architecture.
- Data quality concepts for analytics and ML: discussion of data characteristics that create quality challenges, and considerations specific to ML and analytics workflows.
- Data quality concept framework: conceptual elements including data quality management, data quality governance, and data provenance as they apply to analytics and ML.
- Data life cycle (DLC) for analytics and ML: lifecycle model, stages from conception through discontinuation, and cross-stage processes relevant to data quality.
- Examples and scenarios: informative annex with use cases to illustrate application of concepts in real projects.
This part does not prescribe detailed measurement methods or prescriptive processes; those are addressed in subsequent parts of the series (e.g., models, measures, requirements, governance and visualization).
Practical applications and users
ISO/IEC 5259-1 is intended for stakeholders who need a common conceptual foundation for data quality in AI/ML contexts:
- Data scientists and ML engineers - to align expectations about data characteristics, provenance and lifecycle considerations.
- Data stewards and data governance leads - to frame governance, stewardship and provenance requirements.
- AI project managers and architects - to design processes and data architectures that meet recognized data quality concepts.
- Compliance, risk and audit teams - to interpret terminology and examples when assessing data quality controls and documentation.
- Tool and platform vendors - to ensure features (provenance tracking, data lineage, lifecycle support) align with recognized concepts.
Use cases include preparing training and evaluation datasets, designing data-sharing agreements, implementing data lineage/provenance, and integrating data quality thinking across ML life cycles.
Related standards (ISO/IEC 5259 series)
- ISO/IEC 5259-2 - data quality model, measures and reporting (builds on ISO 8000 / ISO/IEC 25012/25024)
- ISO/IEC 5259-3 - requirements and guidance for establishing and improving data quality processes
- ISO/IEC 5259-4 - organizational approaches for quality of training and evaluation data
- ISO/IEC 5259-5 - data quality governance framework
- ISO/IEC TR 5259-6 - visualization framework for data quality
Keywords: ISO/IEC 5259-1:2024, data quality, machine learning, analytics, data life cycle, data governance, data provenance, ML projects, ISO standard.
Open preview (PDF)Artificial intelligence — Data quality for analytics and machine learning (ML) — Part 1: Overview, terminology, and examples