Any unintended modification, damage or inconsistency in data content or structure that impairs integrity, correctness or usability of records or datasets, arising from hardware faults, software bugs, human error, transmission problems or malicious actions.

Principle

Data integrity is maintained by detection, containment and recovery controls (validation rules, transactions, checksums, access controls, backups and logging); without these, corrupted data can propagate, produce incorrect results and be difficult to isolate or repair.

Demonstration

Illustrative scenario — Situation: A database write is interrupted by a power failure. Recognition: Referential constraints report inconsistent rows. Action: Database administrators roll back the incomplete transaction or restore from a recent transaction log and apply integrity checks before returning the dataset to production. Consequence: If not detected and corrected, downstream users may consume incorrect data and automated processes may act on flawed inputs.

Misapplication

Labeling deliberate or authorized edits, format transformations, or required data normalization as 'corruption'; the semantic error is failing to distinguish unintended integrity loss from intentional, documented change.

Consequence

Data corruption leads to unreliable analyses, operational errors, incorrect decision-making, possible compliance breaches and costs for detection, correction and remediation; corruption can cascade across integrated systems if not contained.

Reversal

A data modification is not corruption when it is intentional, authorized, and recorded with provenance; similarly, systems using strong transactional guarantees and validation will prevent transient faults from producing persistent corruption.

Boundary

Clearly within: unintended file truncation or unintended alteration of database rows that violate constraints. Boundary case: ambiguous records altered by poorly documented batch processes where intent is unclear. Clearly outside: documented schema migrations, audited corrections, or designed transformations.

Semantic Tension

Integrity versus Availability/Performance — mechanisms that enforce strict integrity (synchronous replication, transactional guarantees) can reduce availability or increase latency, requiring trade-offs based on risk tolerance.

Synthesis

Data Corruption frames preservation and operational reliability as a lifecycle problem: prevent by design, detect through validation and monitoring, contain by isolation and rollback, and recover via versioned backups and verified restorations.