Glossary

Every filter on the catalog is built from one of eight dimensions. Filtering is OR within a dimension and AND across them: picking PDF and Image widens the result, then adding Self-hosted narrows it.

A tool must have a value recorded on media, deployment and method before it can be listed at all. The other five are optional and often sparse.

Redaction, and what it is not

Redaction is the permanent removal of information from a document, such that it cannot be recovered from the file.

Masking covers information up. A black rectangle drawn over a paragraph in a PDF viewer hides it from a reader and leaves the text in the file, where anyone can select, copy or extract it. This is the single most common way a "redacted" document leaks, and it is why true content removal is tracked as its own capability below.

Media

What the tool can redact.

• PDF — the default case: documents with a text layer, and scans without one.
• Image — JPEG, PNG and similar, where there is no text layer to remove.
• Video — moving footage, usually faces, licence plates and screens.
• Audio — spoken content, typically bleeped or silenced.
• Text and logs — plain text, CSV, JSON, application logs.

Deployment

Where the tool runs, which is usually the first question a regulated buyer asks.

• Online — a hosted web application. Your documents go to the vendor.
• Desktop — installed software; files stay on the machine.
• Self-hosted — you run it on your own infrastructure.
• API — programmatic access, for redaction inside your own pipeline.
• Browser extension — runs inside the browser against what is on screen.

Redaction method

How the tool decides what to cover.

• Manual — a person selects each region. Slow and fully accountable.
• Rule-based — patterns and dictionaries: card numbers, national ID formats, a list of names.
• AI-powered — a model identifies entities, faces or speech.
• Hybrid — automated detection with a human review step. The common shape for high-stakes work.

Pricing model

How the tool charges. The difference between these is covered at length in pricing units.

Free · Freemium · Subscription · Pay per page · Pay per document · Pay per minute · One-time purchase · Open source · Quote only

Compliance

Regimes and certifications a vendor claims. These are vendor claims recorded as claims — we note what is published, we do not audit it.

HIPAA · GDPR · CCPA · SOC 2 · ISO 27001 · CJIS · FERPA

Platform

Operating systems and runtimes: Windows · macOS · Linux · iOS · Android · Web

Capability

Specific features, and the dimension worth reading closely.

• True content removal — the underlying data is destroyed, not covered. See above.
• OCR — reads text out of scans, without which a scanned PDF cannot be searched for anything to redact.
• Batch processing — many files in one operation.
• Metadata removal — strips author, revision history and EXIF, which survive a visual redaction untouched.
• Audit log — a record of who redacted what and when.
• Face detection, Licence plate detection — automatic detection in images and video.
• Entity detection — names, addresses, account numbers and similar in text.
• Speech to text — transcription, a prerequisite for redacting audio by content.
• Webhooks, SSO, Air-gapped — integration and deployment constraints.

Audience

Who the vendor sells to: Legal · Healthcare · Government · Law enforcement, among others. A tool filed under an audience is one whose own marketing targets it, not one we have judged suitable for it.