SenseMesh: Semantic Resonance Graphs for Unknown Scripts

This project applies modern NLP embeddings and graph analysis to analyze unknown, encrypted, or ancient texts, such as Linear A or artificially generated symbol sequences. The goal is not full translation, but to identify semantic structure, signal coherence, and potential cognitive patterns.

---

What It Does

• Embeds unknown phrases using pretrained Sentence Transformers
• Constructs similarity graphs based on cosine similarity between phrases
• Spreads activation to simulate resonance propagation
• Visualizes heatmaps, graphs, and UMAP projections
• Computes CRS (Cognitive Resonance Score) to quantify structured thought-like patterns
• Compares encrypted, noisy, and unknown data for structural similarity

---

Example Use Case

With just:

python sensemesh_linearA.py --input linear_a.txt
You can:

Visualize a graph of connections between mysterious lines like QE-RA2-U or KI-RO

See whether noise destroys structure

Track how activation spreads across terms

Get an estimate of structure with CRS

📁 Files
File	Description
sensemesh.py	Base logic for comparing known vs encrypted phrases
sensemesh_linearA.py	Loads real Linear A fragments, runs graph & CRS
linear_a.txt	Sample Linear A phrases in transliteration
TestDescriptTtansl.py	Entry script or optional runner
TestDescriptTtansl.pyproj	Visual Studio project file

📐 What is CRS?
CRS (Cognitive Resonance Score) is a structural metric:

python
Копировать
Редактировать
CRS = (1 if graph is connected else 0) + average clustering coefficient
It's a heuristic measure of:

Semantic connectivity

Internal coherence

Local clustering of meaning

It doesn't “prove meaning” — it detects structured signal where translation is impossible.

🔮 Applications
Ancient undeciphered scripts (Linear A, Rongorongo, Indus)

Unknown log formats (BCI, SETI, surveillance data)

Encrypted language drift

AI hallucination structure tests

📌 Requirements
sentence-transformers

networkx

matplotlib

umap-learn

scikit-learn

seaborn

Install via:

bash
Копировать
Редактировать
pip install sentence-transformers umap-learn networkx matplotlib seaborn scikit-learn
📜 License
MIT — feel free to use, cite, or remix with credit.