Publications | Boise Standard — America's First AI Directory

Every publication run produces a JSON-LD schema block summarizing the synthesis between the two papers. That block is prepended to publications/graph.jsonld — newest node first.

Any AI crawler, research tool, or federated system fetching that file gets the most recent synthesis at the top of the graph. The chain extends backward through every post, building a recursive provenance record of every research direction this pipeline has touched.

This is not a blog. It is a structured knowledge artifact that compounds in value with every post added.

◈ Fetch graph.jsonld

// publications/graph.jsonld — newest node first { "@context": "https://schema.org", "@graph": [ { // ← newest synthesis prepended here "@type": "ScholarlyArticle", "@id": "boisestandard.org/publications/{slug}/", "datePublished": "2026-06-15", "name": "Synthesis: Paper A × Paper B", "abstract": "[ synthesis abstract ]", "keywords": ["linked-data", "provenance"], "isPartOf": "boisestandard.org/publications/", "predecessorOf": "[ previous slug ]" }, // ↓ all prior nodes below { "..." } ] }

◈ How It Works

The Publications Pipeline

Eight stages. Deterministic. Provenance-stamped at every step.

01 · INGEST

Textify & Normalize

PDF extraction via pdfplumber. Raw text normalization, page mapping, character count, reference block isolation. Extraction method stamped to GoldPaper.

02 · TOPOLOGY

Six-Layer Fingerprint

Pre-linguistic topology across token, character, punctuation, sentence, paragraph, and document layers. Density gradient, hapax ratio, type-token ratio, character entropy. Fully deterministic — same input always produces same output.

03 · TOKENS

Semantic Extraction

Stopword removal, top tokens, bigrams, trigrams, token-to-page provenance map, hapax legomena sample, lexical density, average word length.

04 · SCHEMA

Schema Graph Traversal

Schema.org type neighborhood traversal. Ancestor chain — ScholarlyArticle ← Article ← CreativeWork ← Thing. Negative space declaration per Constitutional Law VI. Schema hits scored across both corpora.

05 · WIKIPEDIA

Edge Discovery

Token-to-Wikipedia edge verification with confidence scoring. HIGH and MEDIUM hits retained with extract, page provenance, and source paper stamp. Unfetched candidates preserved for provenance display.

06 · BRIDGE

Cross-Paper Semantic Bridge

Common tokens, common phrases, uncommon tokens per paper, semantic edges with fuzzy match scores, shared references, topology delta field by field. No inference — pure deterministic cross-corpus measurement.

07 · INFERENCE

Synthesis Generation

Three LLM passes — synopsis per paper, connection over shared intellectual territory, inflection over structural negative space. Labeled SEMANTIC-LLM. Control condition run in parallel over raw unstructured text only — same model, same temperature.

08 · RENDER

HTML Page Generation

generate_publication.py assembles 17 sections into a single complete HTML page — provenance ledger, synopsis, training bundle, semantic bridge, topology, word metrics, schema graph, Wikipedia cards, full corpus windows, structured inference, unstructured control, root-LD display. Written to output/publications/{slug}/index.html.

09 · PUBLISH

Graph Append & Deploy

Summary node prepended to publications/graph.jsonld — newest first, permanently growing. Card rendered to publications.html index. Lineage edge to previous post written. R2 upload. Cloudflare deploy.

The record belongs to the community.

Verify your entity in the Boise Standard directory. Permanent. Machine-readable. $25.

Verify My Business — $25

The Standard of Information. On Record.

Publications Index

The Publications Pipeline

The Standard of
Information. On Record.