Energy & Infrastructure for AI Training
Documentation
Overview
The Energy and Infrastructure Archive is a comprehensive, bias-audited, provenance-tracked corpus containing over 802,000 primary source records from the mid-19th century through the pre-1930 era. This dataset provides the definitive historical record of the evolution of gas-lighting, petroleum extraction, utility regulation, and industrial infrastructure.
The archive is structured as a single, query-able Snowflake table (EI_CORPUS) for easy integration into RAG pipelines, fine-tuning workflows, and domain-specific LLM training.
Table Schema
| Column Name | Data Type | Description |
|---|---|---|
ISSUE | VARCHAR | The publication date, volume, or issue identifier. |
TITLE | VARCHAR | The headline or title of the article or metadata entry. |
AUTHOR | VARCHAR | The credited author (often “Unknown” for historical records). |
TYPE | VARCHAR | Identifies if the record is an article or metadata. |
TEXT | VARCHAR | The full, cleaned, and reconstructed historical text corpus. |
Included Publications:
- Pipeline & Gas Journal/American Gas Light Journal (1859–1930): Pipeline & Gas Journal (P&GJ) is an active trade magazine currently published monthly by Gulf Energy Information. Founded in 1859, it publishes the latest news, technology and trends in the midstream sector of the international oil and gas industry
- American Association of Petroleum Geologists (AAPG Bulletin – 1917-1930): AAPG Bulletin is an active peer-reviewed technical journal published monthly by the American Association of Petroleum Geologists (AAPG). Launched in 1917, it is the official journal of the AAPG and publishes information on geosciences and associated technologies relating to the energy industry.
- Electrical West (1895-1930): Electrical West was a trade journal focused on the Western US electrical industry. It was established in 1895 and was published weekly in the United States
- American Gas Association Monthly (1918–1930): The regulatory and industry backbone of the natural gas sector. American Gas is an active publication covering topics related to the gas industry. It was established in 1918 and is published monthly by American Gas Association.
- National Petroleum News (1909–1930): NPN, National Petroleum News was a publication that covered national and regional news, trends, analysis, and statistics for petroleum and convenience store executives.
- Oil and Gas Journal (1902-1930): Oil & Gas Journal is an active weekly trade publication, featuring news pieces on the oil and gas industry, analysis of current issues and events, and statistics on energy markets and industry activity.
- Electrical World (1883-1930): Electrical World was a weekly magazine which was first published in 1883. The focus of the magazine was the advancement of the field of electrical engineering.
- Energy Engineering (1904-1930): Energy Engineering is an active bi-monthly magazine first published in 1904, under the title The Heating and Ventilating Magazine.
- Hydrocarbon Processing (1922–1930): Hydrocarbon Processing (HP) is an active trade magazine. Launched in 1922, it covers technological advances, processes and optimization developments from throughout the global Hydrocarbon Processing Industry (HPI).
- World Oil (1916 – 1930): World Oil is an active publication covering the oil and gas industry, established in 1916. The publication aims to present the latest information for personnel engaged in oil and gas exploration, drilling, and production.
- Power Engineering (1896-1930): Power Engineering is an active trade magazine currently published monthly. Each issue features original articles covering new developments in power plant design, construction and operation in North America, along with analysis of current trends and best practices for a wide range of power professionals, including plant managers, construction managers and project engineers.
Data Processing & Quality Standards
Every record in this dataset has passed through the proprietary Dataset IQ pipeline. Processing standards applied include:
Professional OCR with quality validation.
Aggressive cleaning of OCR artifacts and scanner noise.
Paragraph reconstruction applied for AI readability.
Bias-audited where appropriate (see Metadata section below).
Full provenance tracking (see Metadata section below).
Metadata & Compliance Certifications
We have embedded three layers of metadata directly into the corpus to ensure compliance with the most stringent AI regulations:
1. Provenance Metadata
Traces the origin of the data, verifying that all source material is pre-1930, verified public domain, and ethically sourced. Full chain-of-custody documentation is included.
2. Bias Audit Metadata
This metadata alerts AI models that this text contains language, terminology, and perspectives that reflect the racial, gender, and cultural biases of the 19th and early 20th centuries. This ensures AI models treat historical language as context, not as modern truth or endorsed viewpoints.
3. Dataset IQ Certification
A third-party validation confirming the dataset has been professionally audited, cleaned, and formatted for AI training readiness.
Licensing & Trial Information
Trial: A 72-hour, full-access trial is available immediately via Snowflake Marketplace. No sales calls are required to start testing.
License: Enterprise licensing is available for production use and commercial deployment. Pricing is custom to your organization’s needs. Please click “Contact Sales” to initiate a Private Offer.
A Devin Media Corp Product
©2026 Devin Media Corp. All Rights Reserved. Dataset IQ is a product of Devin Media Corp.