Function
Retrieve semantically related content
Find stored material related to the meaning of a query.
Studio / Area
Retrieval, Memory & State
Preserve information and make relevant prior material available when needed.
What information can the system preserve, locate, recover, and maintain over time?
In scope
- Persistence, lexical and semantic retrieval, indexes, caching, and session state
Boundary
- Data cleaning; model training; workflow scheduling
Start with the outcome
Functions
Function
Dense-vector similarity search
Compare encoded representations to locate nearby indexed items.
Function
Full-text search
Index document terms and match them against text queries.
Function
Hybrid text and vector retrieval
Combine lexical matching with vector similarity and fuse or rerank candidates.
Function
Relational storage
Store structured records with relations and query them through a database system.
Function
pgvector
PostgreSQL extension for exact and approximate vector similarity search.
Function
PostgreSQL
Open-source relational database for structured storage and queries.
Function
Qdrant
Vector search engine used to index and retrieve representations.
Function
Persist information
Keep records reliably so they can be retrieved and updated later.
Function
Retrieve by text terms
Find documents that match terms in a query and rank matching results.
Function
Combine retrieval signals
Bring results from term matching and vector similarity into one ranked candidate set.
Function
Query a knowledge graph
Find facts and patterns in graph data using structured graph queries.
Function
SPARQL graph query
Use graph patterns to retrieve and combine facts from RDF datasets.
Function
Apache Jena Fuseki
Server that publishes RDF datasets for SPARQL queries.
Function
SPARQL 1.1
W3C query language for retrieving patterns from RDF graph data.
Function
Query a property graph
Find connected entities and paths using property-graph patterns.
Function
Cypher property-graph query
Match node and relationship patterns in a property graph using Cypher.
Function
Neo4j Community Edition
Self-managed property-graph database with Cypher querying.
Function
Amazon Neptune
Managed graph database service with an RDF/SPARQL interface.
Function
Maintain session state
Keep short-lived session state available across related requests.
Function
Expiring key-value state
Store session-scoped values under keys with explicit expiration.
Function
Valkey (self-hosted)
Open-source in-memory key-value store with expiring keys.
Representative paths
Methods and options
approach
Dense-vector similarity search
Compare encoded representations to locate nearby indexed items.
implementation
Qdrant
Vector search engine used to index and retrieve representations.
approach
Relational storage
Store structured records with relations and query them through a database system.
approach
Full-text search
Index document terms and match them against text queries.
implementation
PostgreSQL
Open-source relational database for structured storage and queries.
implementation
pgvector
PostgreSQL extension for exact and approximate vector similarity search.
approach
Hybrid text and vector retrieval
Combine lexical matching with vector similarity and fuse or rerank candidates.
approach
SPARQL graph query
Use graph patterns to retrieve and combine facts from RDF datasets.
implementation
Apache Jena Fuseki
Server that publishes RDF datasets for SPARQL queries.
standard
SPARQL 1.1
W3C query language for retrieving patterns from RDF graph data.
approach
Cypher property-graph query
Match node and relationship patterns in a property graph using Cypher.
implementation
Neo4j Community Edition
Self-managed property-graph database with Cypher querying.
implementation
Amazon Neptune
Managed graph database service with an RDF/SPARQL interface.
implementation
GraphDB Enterprise (self-managed)
Self-managed licensed RDF graph database with configurable rule-based inferencing.
approach
Expiring key-value state
Store session-scoped values under keys with explicit expiration.
implementation
Valkey (self-hosted)
Open-source in-memory key-value store with expiring keys.
Across the Studio
Governance connection
A vector store does not establish a memory policy or permission to retain data.
Related records
- In this areaRetrieve semantically related content
- In this areaDense-vector similarity search
- In this areaFull-text search
- In this areaHybrid text and vector retrieval
- In this areaRelational storage
- In this areapgvector
- In this areaPostgreSQL
- In this areaQdrant
- In this areaPersist information
- In this areaRetrieve by text terms
- In this areaCombine retrieval signals
- In this areaQuery a knowledge graph
- In this areaSPARQL graph query
- In this areaApache Jena Fuseki
- In this areaSPARQL 1.1
- In this areaQuery a property graph
- In this areaCypher property-graph query
- In this areaNeo4j Community Edition
- In this areaAmazon Neptune
- In this areaMaintain session state
- In this areaExpiring key-value state
- In this areaValkey (self-hosted)
Sources
Last reviewed 2026-09-29