Search Domain

Overview#

An analyst working a people-trafficking investigation types a partial surname into the platform search bar. In under a second, results come back from entity profiles, investigation records, and evidence metadata: a person record matching the name, two documents mentioning it, and a location record three degrees away in the entity graph. She narrows the results with a geographic filter to a specific city and a time window covering the last six months. The Search domain makes this possible by maintaining indexes across the platform's data, combining full-text, semantic, faceted, geospatial, and graph-based search into a single interface with relevance tuning and health monitoring.

Key Features#

• Multiple Search Types: Choose from full-text keyword search, semantic similarity search, faceted filtering, geospatial proximity search, temporal range search, and graph relationship search depending on the nature of the query.

• Index Management: Create and manage search indexes with configurable field definitions, relevance weighting, and analysis settings to optimise search results for specific use cases.

• Index Templates: Define reusable index configurations with preset field mappings, analysers, and scaling parameters for consistent and rapid index creation across the platform.

• Indexing Job Management: Run full rebuild, incremental update, optimisation, and validation jobs to keep search indexes current and performant as data changes.

• Search Analytics: Monitor query patterns, search performance, and index utilisation to understand how users search and identify opportunities to improve search relevance.

• Health Monitoring: Track index health scores, identify issues, and receive optimisation suggestions to maintain reliable search performance.

Use Cases#

Cross-data-type search capabilities apply wherever analysts must move quickly across large, heterogeneous datasets. Relevant industries include law enforcement, financial intelligence, and defence and national security.

• Investigation Search: Search across investigation records, evidence, and associated entities to quickly find relevant information during active cases.

• Entity Discovery: Locate persons, organisations, vehicles, and other entities across the platform using name, identifier, or attribute-based search.

• Geographic Search: Find records within a specified geographic area or radius for location-based investigations and operational awareness.

• Temporal Analysis: Search for events, communications, and records within specific time windows to build investigative timelines.

Integration#

The Search domain connects with other platform capabilities for comprehensive data discovery:

• Investigation Management: Investigation data is indexed for rapid retrieval
• Profile Management: Entity profiles are searchable across all profile types
• Evidence Management: Evidence metadata is indexed for discovery
• Analytics: Search usage patterns feed into operational analytics

Open Standards#

• GraphQL (June 2018 specification): the entire search API, including index management, document indexing, hybrid search, and analytics queries, is exposed through a strongly typed GraphQL schema with cursor-based pagination.
• Okapi BM25: full-text relevance ranking uses PostgreSQL ts_rank with tsvector/tsquery, which implements the BM25 probabilistic term-frequency model to score keyword search results.
• Reciprocal Rank Fusion (RRF, Cormack et al. SIGIR 2009, DOI 10.1145/1571941.1572114): hybrid search fuses BM25 keyword rankings and dense vector rankings using the RRF algorithm with the standard k=60 smoothing constant.
• Dense vector similarity search (ANN cosine similarity): the semantic index type stores and queries float embeddings via approximate nearest-neighbour search, following the standard cosine-similarity scoring model used in bi-encoder retrieval.
• JSON (RFC 8259): document metadata, index configuration, job logs, and analytics payloads are all stored and transmitted as JSON objects.
• ISO 8601: all timestamps for indexes, indexing jobs, documents, and analytics trends are serialised in ISO 8601 extended date-time format.
• OAuth 2.0 / JWT (RFC 6749, RFC 7519): every query and mutation is gated by an IsAuthenticated permission class that validates a JWT bearer credential issued by the platform authorisation server.

Last Reviewed: 2026-02-09 Last Updated: 2026-04-14