PHILIA TextAnalyzer CH7-V5: Evidence Retrieval with Dual-Pool Fetch, CPU Anchor, and Decidability Basis Labeling — 57-Item Full Corpus Confirmatory Pre-registration

This pre-registration declares the confirmatory study design for PHILIA TextAnalyzer CH7-V5 (V1h10), applying a frozen evidence retrieval pipeline to a 57-item claim-verification corpus. The V1h10 pipeline combines dual-pool Wikipedia fetch (subject + object pages), CPU-based anchor prepending for missing slot detection, and Stage1b basis labeling (STATED / DERIVED / ABSENT) as an error visibility layer. The establishment condition (C1) is ≥49/57 = 86.0% Strict X retrieval rate, determined by 4-rater independent re-scoring of the V1h10 exploration run results. This document is part of the PHILIA-EcoSensory Swarm research project (Trinity AI Research Team, Konyang University). Prior confirmatory DOI: 10.5281/zenodo.21988826 (111th, CH7-v30 ESTABLISHED).