Methodology summary
Fact
A value supported by the published record.
Source
Evidence that establishes that fact and its authority.
Public link
A recorded page or document readers can inspect.
How information is verified
Verification is a sequence of evidence and consistency checks, not a promise that every public record has been found. The website publishes only the reviewed result of this workflow.
- 1Locate a public sourceStart with a relevant public page, document, archive, or dataset and identify who published it.
- 2Preserve source identityKeep the publisher, source authority, reference period, and public link attached to the record.
- 3Extract supported factsRecord only the fields the source actually establishes; unrelated claims do not inherit its authority.
- 4Normalize without changing meaningRepresent dates, identifiers, amounts, places, and documentary states consistently while retaining their original meaning.
- 5Validate relationships and totalsCheck identities, references, joins, expected totals, and cross-dataset consistency before publication.
- 6Keep uncertainty visibleLeave unsupported values unknown, preserve meaningful distinctions, and state the dataset’s coverage.
- 7Publish reviewed fields onlyMove only approved, frontend-safe fields into a versioned public export used by the website.
- 8Keep the trust path openWhere available, retain a public link so readers can inspect the source behind a published fact.
Normalization without changing meaning
Public records use different formats and terminology. Normalization makes them comparable and searchable while preserving the distinctions needed to interpret them.
- Documentary state
- AWARDED is not CONTRACTED. Each state describes the strongest published documentary evidence, not physical progress.
- Financial meaning
- ABC is not a winning bid, contract amount, or actual expenditure. These values describe different stages and remain separate.
- Missing values
- Unknown is not zero. An unavailable value stays absent or is shown as “Not specified.”
- Place
- Missing barangay attribution is not converted into an invented location or map point.
- Identifiers
- Project IDs, APP codes, procurement references, PhilGEPS references, contract numbers, and source identifiers retain separate namespaces.
Evidence supports bounded claims
A source supports only the facts it actually establishes. An award document may support an award decision, bidder, or winning bid. It does not automatically prove contract execution, a Notice to Proceed, completion, physical progress, or actual expenditure.
Missing evidence is not proof that an event never happened. It means the current public dataset does not contain qualifying evidence for that claim.
The detailed project-specific lifecycle, amount, identifier, and linkage rules remain on the Project Methodology page. Readers can inspect individual records through Project Sources.
Privacy and the publication boundary
The public website consumes only reviewed, versioned, frontend-safe exports. Research material may include raw sources, working notes, unresolved records, or information needing additional verification or privacy review. Its existence does not authorize publication.
Public frontend export
Reviewed fields, validated relationships, documented coverage, and public source links appropriate for the website.
Not automatically public
Unresolved, sensitive, or unreviewed information stays outside the frontend until verification, minimization, and publication decisions support release.
A publicly accessible source does not mean every field should be republished or aggregated. Person-level information may require stronger minimization, maintenance, and publication decisions. BetterSanFernando does not claim that this conservative review is a formal legal compliance program.
Versioned public releases
The current website reads source data release v0.1.1 through public export contract v0.1.0. The public manifest declares 22 dataset files grouped into 12 published domains.
Versioning makes changes deliberate: the website can validate the expected release, reject unexpected structures, reproduce what a release contained, and avoid silent changes to field meaning. No release date is shown because the current public manifest does not expose one.
Structured exports are checked for valid identifiers, project-to-evidence relationships, reconciled population totals, PSGC consistency, valid source relationships, and manifest integrity before use. The public page explains the outcome without exposing internal validation values.
Different domains have different limits
- Projects
- A bounded infrastructure and public-works subset. Lifecycle is documentary, not physical progress, and no verified project point coordinates currently exist.
- Population
- The PSA 2024 POPCEN baseline, not a projection or current-year estimate.
- Geography
- City and barangay polygons come from a community-maintained geometry source and are not official PSA shapefiles.
- Legislation
- Archive coverage varies by document type and period. Absence from BetterSanFernando does not prove that a measure does not exist.
- City offices
- The current published directory is bounded coverage, not necessarily a complete organizational chart.
- Full Disclosure
- This is a bounded metadata-only export of 10 Full Disclosure Policy report records (2023-2026) with complete, documented verification already present in this repository as of 2026-09-11. It is not a complete Full Disclosure archive: the canonical inventory separately tracks approximately 200 official source records across 2020-2026, of which only these 13 have individual record-level metadata, and only 10 of those 13 have a verified official URL. No report's financial contents (amounts, line items, disbursements) are extracted, summarized, or published anywhere in this repository.
- City Finances
- City Finances is a selected set of official aggregate reports, not audited financial statements or complete City financial history. Source accounting concepts, funds, periods, cumulative treatment, accounting basis, and units remain distinct and must not be combined without explicit compatibility.
Data limitations
Limitations define what the published information can responsibly support. They are part of the data, not a footnote to it.
- Published datasets are bounded and may not represent all City Government activity or complete historical coverage.
- Source availability, authority, and verification dates vary by domain and record.
- Different datasets may describe different reference periods and should not be treated as one simultaneous snapshot.
- Missing records and unsupported fields remain missing rather than becoming zero, inferred events, invented dates, coordinates, officials, or claims of completeness.
- Some information remains outside the public export because additional verification, privacy review, minimization, or a publication-policy decision is still required.
- Evidence uncertainty is preserved when a public source supports only metadata, a limited stage, or a bounded claim.
BetterSanFernando prefers visible uncertainty over unsupported certainty. If a fact cannot be supported by the current public record, it is omitted or clearly marked unavailable.