Logos Coherence™
LOGOS COHERENCE FRAMEWORK · TECHNICAL REPORT

LCF Research Roadmap

Updated
2026-11-05
Target date
when ready

LCF is being written up as one comprehensive technical report, from which two shorter papers will be drawn: one on the validation mechanism and one on the measurement method used to evaluate it. This page shows where that work stands. Publication is gated on readiness, not on a calendar date.

01

Phase line

Current phase: Drafting

14 of 35 sections are drafted. The foundations and the mechanism are nearly complete. The measurement instrument is next.

  1. PHASE 1

    Draft sections

    Write each section of the report.

  2. PHASE 2

    Assemble

    Combine the sections into one report and reconcile them.

  3. PHASE 3

    Review

    Systematic prior-art search, a full independent read, and collaborator review.

  4. PHASE 4

    Publish

    Preprint on an open repository and arXiv, with the open-source code release.

  5. PHASE 5

    Companion papers

    A mechanism paper and a measurement paper, drawn from the report.

02

Section progress

DraftedTo compile from existing workTo write
14 drafted13 to compile8 to write35 sections

Front matter

0 / 3
Title and abstract
Reading guide
Provenance and claim discipline

Foundations

8 / 9

Shared by both companion papers.

The problem: the extraction seam
Thesis and category
Scope of claims
Evaluation corpus
Reference labels and how they were built
Empirical record across six versions
Substrate enrichment arc
Adopter toolkit
Discussion and related work

Part I · The mechanism

5 / 6

How LCF validates language-model output against a semantic substrate.

Architecture
The confirmed-graph boundary
Deterministic verdict path and audit trail
Kernel rules and substrate enrichment
Remediation case study
Limitations and threats

Part II · The instrument

0 / 7

How the errors are measured and attributed, drawing on survey-error methodology.

Survey-error lineage and three-stage decomposition
Three-classification measurement design
Confusion-matrix stage signatures
Specification gaps confirmed by construction
Extraction reliability analysis
Imperfect reference labels
Guarded statistical inference

Appendices

1 / 10
Full corpus table
Reference-label tables
Per-version results
Rater protocol materials
Substrate of record
Figures
Run-artifact provenance index
Adopter recipe
Survey statistics and machine learning concordance
Glossary

“Drafted” means a reviewed draft of the section exists. “To compile” means the material exists and needs assembling into report form. Every section is revised again when the report is assembled and reviewed. Nothing here is a claim about results; results appear only in the published report.