Methodology
How the episodic chronicle is turned into a structured, auditable knowledge graph — and how every claim carries a citation back to its source volume and page.
Overview
Every record, event, and entity in the Compendium carries a citation back to its source volume and page, and — where extraction is uncertain — a confidence score. The structure mirrors the chronicle itself: episodes that gather the detail events within them, each linked to the people and places it names.
Extraction & Structure
Records are extracted from the source volumes — including page-image vision extraction for the Arabic editions, whose text layers are not machine-readable — then segmented by reference number into individual records. In the English editions, events are recognised at two layers (episodes and the detail events within them) under a historical taxonomy, dated to the precision the text supports, and linked to their participants and places; entities are canonicalised and linked across the chronicle. The Arabic editions are, so far, vision-extracted, segmented, and cited only — their entity, event, and place linking has not yet been built. Every record retains a citation to its source volume and page so claims can be traced back to the original.
Why It Matters
Historical scholarship rewards epistemic honesty. By exposing confidence scores, evidence bases, and review trails for every claim, the Atlas lets researchers reason about not only what the corpus says, but how reliable each extraction is and what evidence supports it. Nothing is auto-merged on a hunch; nothing is hidden; and every reviewer decision is durable across re-runs.