Back to glossary

Data quality

Veri kalitesi

D1

Data quality is the degree to which data are fit for a defined purpose, considering attributes such as accuracy, completeness, relevance, consistency, reliability, and currency.

Review status: 2027-02-23

Technical explanation

It affects a dataset’s usability and reusability; in AI workflows, poor-quality or unrepresentative data can affect system trustworthiness.

Conceptual boundaries

Data quality is assessed in relation to purpose and context, rather than as a single permanent score.

Provider-neutral example

Before training a routing model, a team can check whether labels are complete, current, and relevant to the queueing decisions it will support.

Limitations

Quality assessment is a series of lifecycle actions, and poor-quality or unrepresentative data can affect AI-system trustworthiness.

Related concepts

Atomic claims and evidence

  1. 1.1NIST SP 1500-18r2, NIST Research Data Framework Version 2.0
    Source
    NIST SP 1500-18r2, NIST Research Data Framework Version 2.0
    Source role
    Authoritative source
    Exact locator
    Data Quality, p. 49
    Supported claim
    Data quality directly affects a dataset’s fitness for purpose, usability, and reusability, and includes attributes such as accuracy, completeness, update status, relevance, consistency, reliability, presentation, and accessibility.
    Last verification
    Review due
    Scope limitation
    Attribute importance depends on the dataset’s purpose and context.
  2. 2.1NIST AI 100-1, Artificial Intelligence Risk Management Framework (AI RMF 1.0)
    Source
    NIST AI 100-1, Artificial Intelligence Risk Management Framework (AI RMF 1.0)
    Source role
    Authoritative source
    Exact locator
    Appendix B, p. 38, data representation and dataset context bullets
    Supported claim
    Harmful bias and other data-quality issues can affect AI-system trustworthiness, and training datasets can become detached from their intended context or stale relative to deployment.
    Last verification
    Review due
    Scope limitation
    These are risk considerations, not proof that every dataset or system has those problems.