Specialist research. Published by Synoptic.
Sources & methodology

Press release ingestion

Press release deduplication and correction handling

Deduplication should reduce repeated notifications without hiding a material change. Keep a release family, a source record and a content version as separate concepts.

Three identities, three different questions

3 records
Source recordHave we received this provider record before?Provider plus documented ID
Content versionDid the entitled content change?Hash of the exact permitted normalized fields
Release familyAre these records about the same announcement?Explicit source relation, or reviewed clustering decision

Do not use the title as the only key

Two different announcements can have the same headline. The same announcement can appear under different punctuation, language or distributor. Use the provider ID for exact replay handling and a separate, reviewable grouping method for possible syndication copies. A correction with the same ID but changed content should create a new version.

A four-case acceptance exercise

Use synthetic input first. Replay the same payload, update its content under the same source ID, deliver another provider’s copy and deliver a translation. Expected behavior: one record for the replay; a new version for the update; linked but distinct provenance for the copy; and language-aware grouping for the translation.

4 records
Exact replayNo new content versionNo repeated alert
Changed content, same source IDAppend version and observation timeEvaluate correction alert policy
Other provider, similar releaseSeparate source record; possible family linkDeduplicate only if grouping is trusted
Withdrawal or correction noticeRetain event and relationUpdate visible status and notify if required

Avoid a misleading event timeline

Keep first receipt time even if a later version changes the displayed publication date. Save the original announcement status rather than rewriting the past. If the provider has no formal withdrawal relation, preserve the notice and send uncertain matches for review instead of silently removing the original.

Source notes

  1. Finnhub official API specification ↗

    Major press releases endpoint, source names, Enterprise full-text statement and majorDevelopment schema.

  2. newsfilter request/response format ↗

    Metadata schema and source-filtered queries; does not establish downstream full-text rights.

The Synoptic workflow

Evaluate a Synoptic release workflow

Start with the sources and fields the monitoring task requires.

Evaluate a Synoptic release workflow