Method

How records are assessed

Every record passes through the same pipeline before publication: normalisation, classification, severity assessment and deduplication.

1. Normalisation

Documents are converted to plain text, headers and formatting artefacts are stripped, and the substantive body is retained in full. Attachments are extracted and treated as their own records where they carry content of their own.

2. Classification

Each record is assigned a category describing what it is about — for example governance, finance, personnel, legal proceedings, or public communication. Categories are descriptive, not accusations.

3. Severity

• Critical — material involving serious harm, violence, criminal conduct, or severe discrimination.
• High — serious misconduct, significant abuse of position, or major financial irregularity.
• Medium — notable but contained issues: questionable conduct, conflicts, procedural manipulation.
• Low — marginal or contextual material.

Severity is not a legal finding. It is an editorial assessment of how much public-interest weight the content carries.

4. Deduplication

Identical or near-identical records are collapsed. Where a single document contains many separate problematic passages, it is published as one entry summarising the document — not as one entry per passage.

5. Personal data

Entries report the presence of personal data — for instance that a record contains contact details or financial identifiers, and how many — alongside the substance of the record itself.

Limitations

• Summaries are machine-assisted and editorially reviewed; they are compressed paraphrases, not verbatim quotations.
• Dates are taken from document metadata where available and may be approximate.
• Absence of an entry does not imply absence of the underlying record.