Core Concepts

Signals

Understanding signal events and sources

What are Signals?

In Semeion, a signal is any piece of content that needs to be analyzed. Signals can come from social media posts, news articles, forum comments, or any other text source.

Signal Structure

Every signal event contains:

  • source - The origin system (e.g., "twitter", "rss", "api")
  • channel - A sub-category within the source
  • content - The actual text content
  • authorId / authorName - Information about the author
  • metadata - Additional context and properties

Ingestion Flow

When a signal is ingested, it passes through the following pipeline:

Signal ingestion pipeline: PII Redaction, Hash, Embed, Cluster, Score, Gate
Complete diagram · Fit to page

Full-size diagram

Signal ingestion pipeline: PII Redaction, Hash, Embed, Cluster, Score, Gate
  1. PII (personally identifiable information) is redacted
  2. Content is hashed for deduplication
  3. An embedding is generated for semantic analysis
  4. The signal is matched to existing clusters or creates a new one
  5. Trust scores are computed
  6. Gate evaluation determines allow, hold, or reject