Interactive Entity Relationship Explorer Across Articles and Clusters

Summary

Introduce an interactive entity explorer that maps people, organizations, places, and policy topics across articles and clusters, enabling users to navigate relationships and discover connected context faster.

Problem

Users can read article summaries and related stories, but they cannot easily traverse cross-article entity relationships (for example, "which agencies are repeatedly linked to this policy topic"). This limits investigative workflows and cross-source analysis.

Proposed Solution

• Build an entity graph layer from extracted article metadata.
• Add explorer views for entity-centric navigation, relationship paths, and temporal co-occurrence.
• Support drill-down from entity nodes into relevant articles/clusters.
• Add filters by source, date range, trust tier, and language.

Acceptance Criteria

• System extracts/stores normalized entities and relationship edges during ingestion/processing.
• New /entities page supports search, graph visualization, and table fallback.
• Entity detail page shows top connections, trend over time, and linked articles.
• Users can filter graph results by date range and source.
• APIs provide paginated access to entities and relationships.

Technical Notes

• Backend:
  • New collections: entities, entity_edges.
  • Endpoints:
    • GET /api/entities
    • GET /api/entities/:id
    • GET /api/entities/:id/relations
    • GET /api/entities/:id/articles
  • Batch job for normalization/merging aliases (e.g., acronym variants).
• Frontend:
  • Entity Explorer page with graph view and list view.
  • Entity detail page with relation strength and timeline widgets.
• Processing:
  • Reuse existing summarization outputs where possible, add fallback extraction pass for missing entities.

Tasks

• Define entity and edge schemas with normalization strategy.
• Implement extraction + alias resolution pipeline.
• Build relationship scoring (co-occurrence + recency weighting).
• Implement entity APIs with filtering and pagination.
• Create frontend graph explorer and detail pages.
• Add deep links from article/cluster pages into entities.
• Add integration tests for extraction and API responses.
• Document graph semantics and known limitations.

Out of Scope (Initial Iteration)

• Manual graph editing by end users.
• Cross-repository/federated entity graph ingestion.