Methods & coverage

How HealthArchive.ca is being built

This page outlines how snapshots in HealthArchive.ca’s current public health web corpus were captured and are preserved, indexed, and replayed. Coverage is bounded to the existing corpus; any future capture or expansion requires a separately authorized data-continuity decision.

Current archive scope

The initial focus is on federal Canadian public health sites whose content directly underpins clinical guidance, surveillance, or high-impact public communication. Examples include:

  • Public Health Agency of Canada (e.g., disease pages, surveillance reports, immunization guidance).
  • Health Canada (e.g., vaccine and drug pages, environmental and product safety information).

Provincial, territorial, and international sources are outside the current scope. Considering any additional source would require a separately authorized data-continuity decision; no expansion is assumed.

Scope is intentionally constrained and defined with explicit inclusion and exclusion rules per source so the project can prioritize reliable provenance over breadth.

The existing corpus includes labeled annual editions and, where applicable, explicitly labeled ad-hoc captures. There is no default future capture cadence: any new annual or ad-hoc campaign requires a separately authorized data-continuity decision.

Capture methods

HealthArchive.ca uses browser-based crawling and standards-based web archive formats (WARCs). At a high level, each capture works as follows:

  1. Seed URLs are defined for each target domain and path, including rules about what to include or exclude.
  2. A browser-based crawler visits in-scope pages, executes JavaScript where needed, and records responses into web archive files.
  3. Responses are stored in archive files alongside metadata such as capture time, HTTP status, and content type.

Captures are stored in WARCs and indexed into a searchable database. The public site replays archived HTML via the backend and, when available, can also offer higher-fidelity browsing through a replay service. Replay fidelity varies by site and content type.

Date range filters in the archive explorer use UTC capture dates.

Storage & replay

The archive relies on dedicated storage for WARC files. When replay is enabled, a replay engine such as pywb can render higher-fidelity historical snapshots in a browser. The goals for replay are:

  • Preserve the original URL structure where possible.
  • Clearly label capture timestamps and make it obvious that the view is archival, not live.
  • Maintain interactive elements (e.g., dashboards) as faithfully as technical constraints allow.

The interface is intentionally conservative: it prioritizes clarity that you are viewing archived content. Some interactive dashboards, embedded visualizations, or third-party assets may not replay perfectly because of JavaScript, API, or hosting constraints.

Change tracking

HealthArchive.ca compares archived captures to highlight text changes between editions. This is designed for auditability and citation, not interpretation.

  • Changes are computed from archived HTML captures.
  • Comparisons are descriptive only (for example: sections added, removed, or updated) and do not provide guidance.
  • Change feeds are edition-aware by default and reflect labeled editions in the existing corpus rather than committing to a future annual cadence.

Limitations and interpretation

  • Not official guidance: Archived content reflects what public sites showed at the time of capture. It may be incomplete, outdated, or superseded, and should not be treated as current clinical guidance or medical advice.
  • Sampling and coverage: Early phases focus on specific high-value domains and paths. Coverage gaps and known limitations will be documented so that “absence” of an archived page is not misinterpreted.
  • Technical artefacts: Some interactive dashboards, embedded visualizations, or third-party assets may not replay perfectly because of JavaScript, API, or hosting constraints.