xanhnon-LLM

NKTg AI for Organizers https://nktg.org

NKTg AI is not a summarization tool. It is a specialized Language Decoding System that uses physical algorithms to measure semantic energy and extract the Core Content — the core logical genetic code of a text.

1. The Nature of Core Content Core Content is the sentence with the highest density of executable actions and the most concrete outcome in a text. If removed, the text loses key information that cannot be inferred from the rest.

It is an original sentence from the source text — not interpreted or altered by subjective reasoning.

1. Philosophy Generative AI Reads and reinterprets a text using the AI's own language.

NKTg AI Measures the energy of each word and sentence to identify core content that already exists. Returns the author's original linguistic genetic code without generating new content.

1. Text Structural Genetic Code AMP (Amplifying) + DAMP (Damping) + STABLE = 100%

AMP Amplifying > 55% Increasing energy — Actions, Assertions, Execution, Results. Text has a clear focal point; Core Content has high reliability.

DAMP Damping > 55% Decreasing energy — Conditions, Context, Risks, Exceptions, Counterarguments. Text tends toward condition analysis or risk assessment.

STABLE AMP ≈ DAMP Balanced state — Technical info, Pure data, Data tables, Descriptive content. A prominent Core Content may not exist.

1. Output Modes 🧠 Left Brain (Extraction) — Distillation

Standard: Retains the most important sentences by the Golden Ratio. Best for quick reading. Condensed: Removes repetitive or semantically similar sentences. Best for long texts. Essence: Converges to a single Core Content sentence. Recommended when AMP > 55%. 🧠 Right Brain (Addition) — Expansion

Refined: Preserves content nucleus with minimum necessary context. Expanded: Core Content combined with surrounding relevant context. Comprehensive (100%): Full text with DAMPING components marked for easy distinction. 5. Value for Experts & Managers Quantifiability: AMP%, DAMP%, Compression%, retained sentences Causality: Core Content often in Action → Result structures Decision-Making: Quickly evaluate whether to read, examine, or act Objectivity: Algorithm-based, not subjective emphasis 6. Value for General Users Read Quickly: Identify the highest-action-density sentence instantly Evaluate Before Reading: Observe AMP/DAMP in round 1 Objective Comparison: Compare Core Content across multiple texts Identify Key Points: Spot conditions, warnings, or risks 7. Limitations NKTg AI does not perform well on handwritten documents — OCR accuracy drops significantly compared to printed text, which may affect extraction quality.

1. NKTg Law — Variable Inertia Algorithm NKTg = f(x, v, m)

p = m × v

NKTg₁ = x × p (Semantic Potential Energy)

NKTg₂ = (dm/dt) × p (Semantic Kinetic Energy)

NKTg(total) = f(NKTg₁, NKTg₂)

Reference DOI: https://doi.org/10.5281/zenodo.20786471

https://doi.org/10.17632/3z9jktjvx3.1

All processing runs directly in the browser via WebAssembly.

Text never leaves the user's device

No server, no cloud, no remote database

History is stored in localStorage — residing entirely on the user's device

© 2026 NKTg AI. All rights reserved.