Tip: select any text on this page to bring up the Ask tooltip and question the Synedrix OS assistant about it.
What Semantica is
Semantica is a Python knowledge-graph and GraphRAG toolkit, real and active at PyPI v0.6.5, bundling several independent capabilities: Docling-based structural document parsing, an in-memory decision graph (ContextGraph), genuine W3C PROV-O export, bridges to several graph databases, and a Plotly-based visualiser.
It also ships a full FastAPI Knowledge Explorer web app and a stdio-only MCP server with 12 tools, but no HTTP/SSE transport, so it must be subprocess-spawned by a client rather than called over a port.
What it is normally used for
How Semantica is used generally, before Synedrix OS enters the picture.
- GraphRAG retrieval: multi-hop questions a pure vector search structurally cannot answer
- Decision and provenance tracking, exported in a standards-compliant format
- Structural document ingestion: extracting tables and layout faithfully before chunking
What it owns, and what it does not
Every platform in the stack has a non-overlapping responsibility. The boundary is as much a part of the design as the capability.
Owns
- Nothing in production: a Semantica-built Qdrant collection and decision graph are both lab-only, additive, and never a replacement for the production KB
Does not own
- Semantic retrieval and similarity search: the single vector store
- → Qdrant (Layer 7); Semantica's collection is additive, never a replacement
- Explicit relationships and lineage as production authority
- → Neo4j (Layer 7); a lab-built graph is an evaluation artifact, not a second production graph store
- Decision provenance already covered by the canonical envelope
- → provenance.source_event_id / derived_from (Memory Fabric Spec §4.2/§4.4)
Why Semantica
Pre-1.0 despite a Production/Stable classifier
Commit history is heavily concentrated (1,823 of the top contributor's commits versus 239 for the next-largest, across 54 nominal contributors), a materially less mature project than its PyPI classifier implies.
Its RDF and graph-algorithm capabilities duplicate what the OS already has
A triple store would duplicate Neo4j, and Semantica's graph algorithms are self-implemented Python, not a Neo4j GDS integration: the same store-duplication rule that already kept EverOS's runtime out, pre-judging those two capabilities out of scope here too.
The one genuinely novel claim has a first signal, not a verdict
Whether Docling's structural parsing measurably improves retrieval over the existing chunker is what LAB-002 measures. The first run moved recall@10 on multi-hop questions from 0.25 to 0.75 at four questions per category; a larger, repeated run decides it.
Licence position
Semantica questions, answered
Short, citable answers drawn from this page.
What role does Semantica play in Synedrix OS?
Semantica (GraphRAG / knowledge-graph toolkit) is mapped to Lab component: no OS layer in Synedrix OS. Structural parsing and decision-graph tooling, evaluated one capability at a time, not one adoption question.
What does Semantica own in Synedrix OS?
Semantica owns: Nothing in production: a Semantica-built Qdrant collection and decision graph are both lab-only, additive, and never a replacement for the production KB.
What does Semantica not own in Synedrix OS?
Semantica does not own: Semantic retrieval and similarity search: the single vector store (owned by Qdrant (Layer 7); Semantica's collection is additive, never a replacement); Explicit relationships and lineage as production authority (owned by Neo4j (Layer 7); a lab-built graph is an evaluation artifact, not a second production graph store); Decision provenance already covered by the canonical envelope (owned by provenance.source_event_id / derived_from (Memory Fabric Spec §4.2/§4.4)).
The full reference page, with the operating model, deployment detail, and links to every canonical specification this platform touches, lives in the documentation repository, which is private while the platform is being built.