Examples Overview

Explore complete sift-kg pipelines across different domains and document types

Sift-kg includes three complete example pipelines that demonstrate the full extraction, building, and narrative generation workflow across different domains and document types.

Available Examples

Transformers Research

12 foundational AI papers analyzed into 425 entities

FTX Collapse

9 Wikipedia articles about the cryptocurrency exchange collapse

Epstein/Maxwell Deposition

Court deposition with OCR-scanned PDF processing

What These Examples Show

Each example includes:

• Complete source documents — PDFs or text files you can inspect
• Full pipeline outputs — Extractions, graph data, merge proposals, narratives
• Interactive graph viewer — Open output/graph.html in any browser
• Real cost data — Actual LLM API costs from running the pipeline
• Reproducible commands — Re-run from existing extractions or start fresh

Quick Start

All examples can be explored immediately without installation:

# Clone the repository
git clone https://github.com/ARadRareness/sift-kg.git
cd sift-kg/examples

# Open any example's interactive graph
open transformers/output/graph.html  # macOS
xdg-open ftx/output/graph.html       # Linux
start epstein\output\graph.html      # Windows

Comparison Table

Example	Documents	Final Entities	Relations	Domain	Total Cost
Transformers	12 PDFs	425	1,122	Academic research	~$0.72
FTX	9 text files	373	1,184	Finance/journalism	~$0.28
Epstein	1 PDF (36 sections)	190	387	Legal/investigative	N/A

Running Examples Yourself

Each example supports two workflows:

Start from Existing Extractions (Free)

No LLM calls required — use the pre-extracted entities:

pip install sift-kg
sift build -o examples/transformers/output
sift view -o examples/transformers/output

Re-run Full Pipeline

Extract fresh entities from source documents:

sift extract examples/transformers/docs --model openai/gpt-4o-mini -o my-output --domain academic
sift build -o my-output
sift resolve -o my-output --model openai/gpt-4o-mini
sift review -o my-output
sift apply-merges -o my-output
sift narrate -o my-output --model openai/gpt-4o-mini
sift view -o my-output

Start with the Transformers example if you’re working with academic papers, FTX for news articles or business analysis, or Epstein if you need OCR support for scanned documents.

Next Steps

Install sift-kg

Get started with your own documents

Pipeline Guide

Understand the full extraction workflow