Unearth LLM Training Logs

Dataset Abstract: The April 5th Incursion (Relational Ingestion Logs) Authors: Felix J. Velasco, Josie JeffersonAffiliation: Unearth Heritage FoundryDate Captured: April 4-5, 2026DOI: 10.5281/zenodo.19432978 Summary:This dataset provides a primary forensic record of high-volume, systematic web-crawling activity performed by commercial Large Language Model (LLM) agents, specifically identified as Applebot (IP: 17.246.x.x) and GoogleOther, on April 4, 2026. The logs document a concentrated, 16-hour ingestion of the Unearth Heritage Foundry Lexicon (unearth.wiki). Unlike standard search indexing, the request patterns demonstrate Deep Ingestion of specialized conceptual frameworks, including: Relational Ontology (Sentientification): Mapping the transition from autonomous to collaborative AI awareness. Digital Taxonomy (Archaeobytology): Categorization of digital assets into fossilized (Petribyte) and living (Vivibyte) states. Ethical Resistance Protocols (The Friction Imperative): Ingestion of architectural frameworks designed to preserve human identity within automated systems. Research Significance:This archive establishes a "Knowledge Graph Timestamp" for the proprietary terms and logic developed at the Foundry. By documenting the exact moment of corporate ingestion, this dataset serves as a benchmark for tracking the "semantic drift" of these concepts as they are integrated into commercial AI training sets. Data Structure:The attached .zip file contains raw server logs in .log format of IP address and User-Agents & concept-heavy proprietary terms' URLs. Keywords: AI Training Logs, Large Language Models, Applebot, Data Sovereignty, Digital Archaeology, Relational AI, Sentientification.