Skip to content

Use Cases: Practical Workflows

This guide catalogs real-world workflows for the Knowledge Graph System. Each use case demonstrates specific techniques for multi-ontology ingestion, semantic analysis, and knowledge extraction.

Available Use Cases

1. GitHub Project History Analysis

Mining your repository for knowledge using GitHub CLI

Extract commit messages and pull requests with gh CLI, organize them into directories, and ingest as separate ontologies. The graph automatically discovers connections between commits and PRs, enabling semantic search across your entire project history.

Key Techniques: - GitHub CLI (gh) for data extraction - Directory-based ontology organization - Multi-source automatic concept linking - Temporal analysis and contributor insights

What You'll Learn: - Why features were implemented certain ways - Who has expertise in specific areas - How architectural decisions evolved - Related commits/PRs (even without explicit references) - Team development patterns

Read full guide →

Planned Use Cases

These use cases are planned for future documentation. Contributions welcome!

2. Research Paper Analysis

Status: Planned

Ingest related research papers as separate ontologies, discover connections between research threads, trace citations and concept evolution.

Status: Planned

Multiple versions of contracts or regulations tracked over time, comparing changes and concept evolution across revisions.

4. Knowledge Base Migration

Status: Planned

Ingest existing documentation sets from different sources, find gaps and redundancies, unify terminology.

5. Meeting Notes and Project Documentation

Status: Planned

Combine meeting transcripts with technical documentation for complete project context, linking discussions to decisions.

6. Multi-Language Code Documentation

Status: Planned

Ingest documentation from different programming language ecosystems, find patterns and translate concepts across languages.

7. Customer Support Knowledge Base

Status: Planned

Build searchable knowledge from support tickets, FAQs, and resolution notes, automatically categorizing issues.

Contributing Use Cases

Have you developed a novel workflow using the Knowledge Graph System? We'd love to include it!

How to Contribute a Use Case

  1. Create the use case document:
  2. Create a new file in docs/guides/use_cases/your_use_case.md

  3. Follow the structure of existing use cases (see github_project_history.md)

  4. Include these sections:

  5. Title and introduction - Problem statement and key insight
  6. Prerequisites - Tools and setup required
  7. Workflow - Step-by-step instructions with commands
  8. What This Enables - Benefits and capabilities unlocked
  9. Tips and Best Practices - Lessons learned
  10. Cost Considerations - Estimation and budgeting
  11. Limitations and Gotchas - Known issues and workarounds

  12. Example queries - Real queries and results

  13. Update this index:

  14. Add your use case to the "Available Use Cases" section above
  15. Include a brief description and key techniques

  16. Link to your detailed guide

  17. Submit a pull request:

  18. Describe the use case and its value
  19. Include any sample data or scripts if helpful
  20. Reference related issues or discussions

Use Case Template

# Your Use Case Title

## [Compelling Subtitle/Hook]

**The Insight:** What problem does this solve? Why is this powerful?

**What You'll Learn:**
- Specific benefit 1
- Specific benefit 2
- ...

**The Approach:** High-level workflow summary

## Prerequisites

Tools, accounts, or setup required

## Workflow

### Step 1: [Action]
Detailed instructions with code examples

### Step 2: [Action]
...

## What This Enables

Specific capabilities and use cases

## Tips and Best Practices

Lessons learned, gotchas, optimization tips

## Cost Considerations

Estimation formulas and budgeting guidance

---

**Last Updated:** YYYY-MM-DD

**Related Documentation:**
- [Relevant guide 1](link)
- [Relevant guide 2](link)

General Workflow Patterns

Across all use cases, these patterns emerge:

Multi-Ontology Organization

Pattern: Organize related but distinct data sources as separate ontologies - Example: Commits vs pull requests, or papers vs patents - Benefit: Targeted querying and clearer data lineage - Automatic linking: Graph connects concepts across ontologies

Directory-Based Ingestion

Pattern: One document per file, organized in directories - Example: project_history/commits/*.txt and project_history/pull_requests/*.txt - Command: kg ingest directory path/to/dir --ontology "name" - Benefit: Simple, file-system-based organization

Metadata-Rich Documents

Pattern: Structure documents with metadata headers 

Title: Document Title
Author: Jane Doe
Date: 2025-10-14
Tags: tag1, tag2, tag3

[Main content...]
- Benefit: LLM extracts structured metadata as concepts - Enables: Author-based queries, temporal analysis, tag-based filtering

Incremental Updates

Pattern: Add new documents to existing ontologies - Deduplication: Graph automatically detects duplicate content via SHA-256 hashing - Growth: Knowledge compounds over time - Benefit: No need to re-ingest entire corpus

Last Updated: 2025-10-14

Related Documentation: - 03-INGESTION.md - Detailed ingestion configuration - 03-EXAMPLES.md - Query examples and results - 02-CLI_USAGE.md - Complete CLI command reference - QUICKSTART.md - Getting started guide