doc-kg 0.12.3

A tool to build a semantically searchable knowledge graph from markdown and text documents

DocKG — A Hybrid Knowledge Graph for Document Corpora with Semantic Indexing and Source-Grounded Passage Packing

Author: Eric G. Suchanek, PhD Flux-Frontiers, Liberty TWP, OH

---

Overview

DocKG constructs a deterministic, explainable knowledge graph from a corpus of Markdown, plain-text, and PDF documents. It semantically chunks text, discovers structural and semantic relationships between sections and chunks, stores them in SQLite, and augments retrieval with vector embeddings via LanceDB.

Structure is treated as ground truth; semantic search is strictly an acceleration layer. The result is a searchable, auditable representation of a document corpus — an ideal retrieval engine for LLMs and a practical foundation for Knowledge-Graph RAG (KRAG).

DocKG uses the same architecture as CodeKG but targets natural-language documents rather than Python source code.

---

Features

• Multi-format ingestion — .md, .txt, .rst, and .pdf (native — no inference)
• Semantic chunking — Heading-structure and paragraph-aware segmentation
• Deterministic knowledge graph — SQLite-backed canonical store with typed nodes and provenance-tracked edges
• Relation extraction — Topics, named entities, and keywords per chunk; co-occurrence and similarity edges built automatically
• Hybrid query model — Semantic seeding (LanceDB embeddings) + structural expansion (graph traversal)
• Passage packing — Context-rich text passages grounded to source documents with headings
• Corpus health analysis — Per-document metrics, hot chunks, orphan detection, coverage report
• Temporal snapshots — Save and diff graph metrics over time
• MCP server — Four tools for AI agent integration (graph_stats, query_docs, pack_docs, get_node)
• Streamlit web app — Interactive graph browser, hybrid query UI, and passage pack explorer

---

Quick Start

# Index a document corpus (SQLite + LanceDB in one step)
dockg build docs/

# Natural-language query — returns ranked document chunks
dockg query "authentication flow"

# Source-grounded passage pack — paste straight into an LLM prompt
dockg pack "configuration reference" --format md --out context.md

---

Installation

Requirements: Python ≥ 3.12, < 3.14

# pip
pip install doc-kg

# With Streamlit web visualizer
pip install 'doc-kg[viz]'

# Poetry
poetry add doc-kg

For advanced deployment options (Streamlit Cloud, Fly.io, offline model cache, git hooks) see docs/deployment.md.

---

Usage

Build and query

dockg build docs/                                    # full pipeline
dockg build docs/ --update                           # incremental (keep existing)
dockg build docs/ --exclude-dir archive              # skip directories
dockg query "deployment configuration"               # hybrid search
dockg pack "error handling" --format md --out ctx.md # passage pack

Analyze corpus health

dockg analyze docs/             # full report + JSON snapshot
dockg analyze docs/ --quiet     # CI mode — exits 1 on issues

Snapshots

dockg snapshot save 0.12.0      # capture current metrics
dockg snapshot diff 0.11.0 0.12.0  # compare two versions

Full flag reference for every command: docs/CLI.md Query patterns and MCP examples: docs/CHEATSHEET.md

---

MCP Integration

Start the MCP server, then wire it into your AI agent:

dockg mcp --repo docs/

Claude Code / Kilo Code — add to .mcp.json:

{
  "mcpServers": {
    "dockg": { "command": "dockg-mcp", "args": ["--repo", "."] }
  }
}

GitHub Copilot — add to .vscode/mcp.json:

{
  "servers": {
    "dockg": { "type": "stdio", "command": "dockg-mcp", "args": ["--repo", "."] }
  }
}

Tool	Description
graph_stats()	Node and edge counts by kind
query_docs(q, k, hop)	Hybrid semantic + structural search
pack_docs(q, k, hop)	Source-grounded passages as Markdown
get_node(node_id)	Fetch a single node by ID

Full provider setup (Claude Desktop, Cline, SSE transport): docs/MCP.md

---

Python API

from doc_kg import DocKG

kg = DocKG(corpus_root="docs/")
kg.build(wipe=True)

result = kg.query("deployment configuration", k=8, hop=1)
for node in result.nodes:
    print(node["id"], node["name"])

pack = kg.pack("authentication flow")
pack.save("context.md")

---

Knowledge Graph Schema

Node kinds

Kind	Description
document	A source .md, .txt, or .pdf file
section	A heading-delimited region within a document
chunk	A semantically coherent text passage
topic	A topic extracted from chunk text
entity	A named entity (person, place, org, concept)
keyword	A keyword or key phrase

Edge types

Type	Description
CONTAINS	Parent → child (document→section, section→chunk)
NEXT	Sequential ordering between same-level nodes
REFERENCES	Chunk references another document or section
SIMILAR_TO	Semantic similarity between chunks (LanceDB-derived)
HAS_TOPIC	Chunk → topic
MENTIONS_ENTITY	Chunk → named entity
HAS_KEYWORD	Chunk → keyword
CO_OCCURS_WITH	Co-occurrence between topics/entities within a chunk

---

Storage Layout

.dockg/
  graph.sqlite      # SQLite knowledge graph (nodes + edges)
  lancedb/          # LanceDB vector index
  snapshots/        # Temporal metric snapshots (JSON)
    manifest.json
    <commit>.json

---

Citation

If you use DocKG in research or a project, please cite it:

APA

Suchanek, E. G. (2026). DocKG: Hybrid Knowledge Graph for Document Corpora (Version 0.12.1) [Software]. Flux-Frontiers. https://doi.org/10.5281/zenodo.19770973

BibTeX

@software{suchanek_doc_kg,
  author    = {Suchanek, Eric G.},
  title     = {{DocKG}: Hybrid Knowledge Graph for Document Corpora},
  version   = {0.12.1},
  year      = {2026},
  publisher = {Flux-Frontiers},
  url       = {https://github.com/Flux-Frontiers/doc_kg},
  doi       = {10.5281/zenodo.19770973},
}

License

Elastic License 2.0 — free for non-commercial and internal use; commercial redistribution requires a license from Flux-Frontiers.