Eclipse RDF4J Documentation

Eclipse RDF4J is a modular Java framework designed for working with RDF (Resource Description Framework) data. It provides a vendor-neutral API for parsing, storing, inferencing, and querying RDF and OWL data.

Key capabilities include:

• SPARQL Support: Full support for SPARQL 1.1 query and update languages.
• Format Support: Support for mainstream RDF formats including RDF/XML, Turtle, N-Triples, N-Quads, JSON-LD, TriG, and TriX.
• Connectivity: Transparent access to both local storage and remote SPARQL endpoints using the same API.
• Extensibility: A modular architecture that allows connecting to various storage solutions and adding functional extensions like full-text search, RDFS inferencing, and SHACL/SPIN validation.

The 3.1.0 release introduced several key improvements for developers:

• Federated Querying: Integrated support for sophisticated federated querying via FedX.
• ElasticSearchStore: A new experimental storage backend for RDF data.
• Improved Iteration: QueryResult and RepositoryResult objects now implement Java Iterable, allowing for simpler client-side iteration.
• Context Stripping Helpers: New helper functions are available to strip context/named graph information from Statement and Model objects.

The 4.0.0 Milestone 1 release includes several performance and usability enhancements:

• SparqlBuilder: Improvements to the API for building SPARQL queries.
• SPARQL Engine: Performance improvements in the query engine.
• SHACL Engine: Performance improvements in the SHACL engine.
• Spring Integration: Improved support for using RDF4J within Spring and Spring Boot applications.

RDF4J 5.3.0 introduces several key improvements for query observability, deployment, and performance:

• Query Observability: Enhanced EXPLAIN support featuring runtime telemetry and a new 'compare mode' in the Workbench.
• Deployment: A new standalone rdf4j-server-boot distribution is available for deploying the Server and Workbench.
• Performance: Optimized workloads involving LMDB, Lucene, and DISTINCT-heavy SPARQL queries.
• Data Export: Direct export of SPARQL results to .xlsx and .ods formats.
• Developer APIs:
  • An experimental query renderer.
  • A new fluent ShaclValidator API for SHACL validation tasks.

RDF4J 2.5.0 introduced several major improvements to the core library and its tooling:

• SHACL Support: Streamlined initialization and an extended range of supported SHACL features.
• RDF4J Console: New commands added to show and edit namespaces.
• RDFS Inferencing: The old RDFS inferencer was replaced with a new, faster Schema-caching RDFS inferencer.
• SparqlBuilder: Various improvements and bug fixes to the SPARQL query construction API.

RDF4J 5.0.0 introduces several major updates and improvements to the framework:

• JSON-LD 1.1 support: Enhanced support for the JSON-LD 1.1 specification.
• FedX Improvements: Various enhancements to the FedX federated query engine.
• SHACL Validation: Improved SHACL validation capabilities, including support for sh:closed and pairwise validation.
• LmdbStore: Stability and performance improvements for the LmdbStore storage engine.
• MapDB Upgrade: Upgraded MapDB, which allows queries with intermediary results to be limited by available disk space rather than being constrained by RAM or Java heap.

RDF4J 4.0.0 Milestone 3 introduces several key improvements and features:

• BCP47 Language Tags: Improved normalization of BCP47 language tags.
• Lucene Full-Text Search: Added support for type/language-based filtering.
• Concurrency: Support for concurrent reads on a single connection.
• SHACL Support: Ability to store SHACL Shapes in user-specified graphs.
• Performance: Various performance improvements across Native, LMDB, and Memory stores.

The Resource Description Framework (RDF) is a standard for expressing information about resources in a machine-processable way. RDF models data as a graph consisting of statements (also called triples).

Each statement consists of three parts:

• Subject: The starting node, representing the resource the fact is about.
• Predicate: The property (edge) that denotes the relation between the subject and the object.
• Object: The end node, representing the value of the property (which can be another resource or a literal).

Resources are globally identified using IRIs (Internationalized Resource Identifiers). Values that are simple strings of characters rather than resource identifiers are called literals.

An ExecPlan is a self-contained design document used to guide a coding agent (or a novice developer) through the implementation of a feature or system change.

Key characteristics of a valid ExecPlan include:

• Self-containment: It must contain all necessary knowledge, context, and instructions so that a user with only the current working tree and the plan can succeed without external documentation or prior repository knowledge.
• Living Document: It must be updated continuously to reflect progress, design decisions, and new discoveries.
• Outcome-Oriented: It focuses on producing demonstrably working behavior (e.g., observable terminal output, HTTP responses, or passing tests) rather than just code changes.
• Plain Language: All technical terms or jargon must be defined in plain language within the plan itself.

When running parallel transactions, concurrent writes can lead to validation failures that wouldn't occur in isolation.

To mitigate this, ShaclSail uses locking when the isolation level is set to SNAPSHOT. This runs transactions sequentially and is typically 2-4x faster than using SERIALIZABLE isolation. Locking only affects transactions writing to the ShaclSail and is applied during commit().

You can disable this specific type of validation using setSerializableValidation(false).