RagLeap Graph Java
A Java implementation of a graph database retrieval library, supporting LLM entity-relation extraction, hybrid vector-plus-graph retrieval, and Cypher query generation, with behavior equivalent to the Python version ragleap-graph.
Timeline (last 90 days)
RagLeap Graph Java v0.2.0 has been released, providing core graph retrieval capabilities, including optional LLM-based entity and relation extraction, configurable extraction and entity deduplication, cross-chunk relation handling, hybrid vector + graph retrieval, pure graph retrieval, chunk citations and graph path citations, pluggable graph and vector search interfaces, backfill tools and a schema migration runner.
The development team verified the Java implementation's behavioral consistency against the Python version ragleap-graph v0.9.0, using recorded calls, parameters, outputs and scripted model responses.
Parity testing covered 422 string similarity pairs, 25 extraction configuration cases, entity and relation extraction, graph upsert (including Cypher statements and parameters), backfill and migration behavior, and retrieval scenarios and retrieval configurations.
The team deliberately injected 27 mutations into the new functionality, and every injected breakage was caught by the parity tests.
The full local run used generated parity fixtures and a real running Neo4j 5.26.29 Community instance, with results: 129 tests, 0 failures, 0 skipped.
The Maven coordinates for RagLeap Graph Java are io.github.antonyrag:ragleap-graph:0.2.0.
The team resolved the Maven artifact in a clean project and ran health checks, upserts, backfills, and pure graph retrieval against a real Neo4j.
The current limitations the team listed include: LLM extraction has not yet been validated against real model providers, compatibility with other Neo4j deployments needs more testing, the optional vector search adapter has not yet been tested against a real RagLeap facade, and the PostgreSQL audit writer has not yet been ported.
All Facts (8)
The current limitations the team listed include: LLM extraction has not yet been validated against real model providers, compatibility with other Neo4j deployments needs more testing, the optional vector search adapter has not yet been tested against a real RagLeap facade, and the PostgreSQL audit writer has not yet been ported.
50%UnverifiedThe development team verified the Java implementation's behavioral consistency against the Python version ragleap-graph v0.9.0, using recorded calls, parameters, outputs and scripted model responses.
50%UnverifiedParity testing covered 422 string similarity pairs, 25 extraction configuration cases, entity and relation extraction, graph upsert (including Cypher statements and parameters), backfill and migration behavior, and retrieval scenarios and retrieval configurations.
50%UnverifiedThe team deliberately injected 27 mutations into the new functionality, and every injected breakage was caught by the parity tests.
50%UnverifiedThe full local run used generated parity fixtures and a real running Neo4j 5.26.29 Community instance, with results: 129 tests, 0 failures, 0 skipped.
50%UnverifiedThe Maven coordinates for RagLeap Graph Java are io.github.antonyrag:ragleap-graph:0.2.0.
50%UnverifiedThe team resolved the Maven artifact in a clean project and ran health checks, upserts, backfills, and pure graph retrieval against a real Neo4j.
50%UnverifiedRagLeap Graph Java v0.2.0 has been released, providing core graph retrieval capabilities, including optional LLM-based entity and relation extraction, configurable extraction and entity deduplication, cross-chunk relation handling, hybrid vector + graph retrieval, pure graph retrieval, chunk citations and graph path citations, pluggable graph and vector search interfaces, backfill tools and a schema migration runner.
50%