shawngraham

Knowledge Graph Extraction for Embedding Models

A pattern for building temporally-grounded knowledge graphs from scholarly and documentary sources, optimised for downstream knowledge graph embedding (KGE) training.

---

The core problem

Most knowledge graph construction from text produces triples: (subject, predicate, object). Applied to scholarly or historical sources, this collapses a critical distinction. A monograph arguing that the Treaty of Westphalia established state sovereignty is doing something categorically different from the same monograph reporting that previous historians have made this argument. The first is a knowledge claim being advanced; the second is a claim being attributed. Embedding models trained on undifferentiated triples will represent both as equivalent facts which could create a fundamental error of provenance.

shawngraham

"""
groq_ocr.py

Processes newspaper images using Groq's vision API and extracts individual
articles to a CSV. Each row represents one article with associated page metadata.

Usage:
    python newspaper_ocr.py --input_dir processed_output/images --output ocr_results.csv

Requirements:

shawngraham

# AUFBAU
## RECONSTRUCTION
**AN AMERICAN WEEKLY PUBLISHED IN NEW YORK**
**by The New World Club, Inc., 209 West 48th Street, New York 19, N. Y. Phone: CIrcle 7-4462**
*Entered as second-class matter January 20, 1934, at the Post Office New York, N. Y. under Act of March 3, 1879*
**Vol. XV—No. 26 | NEW YORK, N. Y., FRIDAY, JULY 1, 1949 | Price 10¢**

***

> **Zunächst in "Aufbau":**

shawngraham

**Role:** You are a precise archaeological document analyst specializing in the digitization of field notebooks and excavation catalogues.

**Task:** 
1. Perform a spatial analysis of the document to distinguish between text blocks, artifact photographs/sketches, and marginalia.
2. Extract metadata and create a brief 2-3 sentence overview of the document's contents.
3. Transcribe the document EXACTLY as written into a valid YAML structure.
4. Extract archaeological entities into specific categories based only on explicit mentions.

**Critical Rules:**
- **Zero Hallucination:** Only include information directly visible in the image. If a word is illegible, mark it as `[illegible]`.

shawngraham

shawngraham

shawngraham

shawngraham

shawngraham

import mesa

class LetterAgent(mesa.Agent):
    def __init__(self, model):
        super().__init__(model)
        self.letters_sent = 0
        self.letters_received = 0
        
    def step(self):
        print(f"Hi, I am agent {self.unique_id}.")

shawngraham

%%capture
!python3 -m pip install git+https://github.com/p-lod/plodlib
!pip install requests_cache
!pip install rdflib

import plodlib
import json
import pandas as pd
from string import Template
import rdflib as rdf