Skip to content

MemoryForgeAILabs/NovaFS

Repository files navigation

NovaFS Symbolic Routing Demo

NovaFS is a proof-of-concept symbolic routing engine. It demonstrates how simple symbolic cues can drive intelligent retrieval and memory systems.

This lean prototype is designed to be easy to understand and extend.

⚠️ Status: Archived (2026). NovaFS is no longer actively developed and is published here as a public reference. The problem it explored — turning natural-language intent into the correct database query or data route deterministically, without embeddings or vector search — is now handled more flexibly by modern natural-language-to-SQL (NL→DB) query models. This repo preserves the symbolic-routing kernel and its glass-box design. See Background & Context and Why this is archived.


🚀 Overview

NovaFS lets you:

✅ Define symbolic cues in cue_map.json
✅ Map those cues to files in symbol_map.json
✅ Route cues to file paths via a simple command-line interface
✅ Log every routing event for transparent traceability

This project can serve as a foundation for more advanced systems in AI memory, search, and knowledge management.


📖 Background & Context

The code in this repository is the kernel of a larger idea. The core premise of NovaFS was deterministic semantic routing:

Map a natural-language query to the correct data — the right file, table, or knowledge category — using fixed, inspectable rules instead of embeddings or vector similarity search.

The toy demo here routes symbolic cues (alpha, 🔺) to files. In its more developed form the same mechanism grew into a semantic abstraction layer that could answer one plain-language question (e.g. "check brake history") across several wildly different database schemas — from free-text notes to fully normalized tables — without the caller needing to know where the data lived.

The design favored a few principles consistently:

  • Deterministic — the same input always resolves to the same route, so behavior is reproducible.
  • Glass-box / auditable — every routing decision is logged with the cue, the chosen path, and a timestamp. No black box.
  • No model in the hot path — routing is keyword/concept matching plus a lightweight reinforcement-and-decay weighting of paths that succeed or fail over time. Fast and cheap.
  • Concept-first — queries route to stable concepts, decoupling user intent from the volatile structure of whatever store sits underneath.

The trade-off was deliberate and, ultimately, decisive: it traded flexibility for predictability. A rule-based router cannot handle ambiguous or novel phrasing the way a learned model can.


🪦 Why this is archived

The specific job NovaFS did well — translating natural language into the right query/route over structured data — is now done directly, and more flexibly, by natural-language-to-database (NL→DB / NL→SQL) query models. Those models close exactly the gap NovaFS's deterministic approach left open: handling ambiguous, unseen, and conversational phrasing without hand-built concept maps and routing tables to maintain.

Given that, the project is no longer worth actively developing. It is kept public as a compact, readable reference for:

  • the symbolic-routing kernel (see routing_engine.py),
  • the glass-box, deterministic design stance, and
  • the path reinforcement/decay weighting idea.

If you're solving the NL→data problem today, reach for an NL→SQL model first. NovaFS is here to illustrate an alternative point in the design space — one that prizes determinism and auditability — not as something to build on.


⚙️ Installation

  1. Clone this repository:

    git clone https://github.com/MemoryForgeAILabs/NovaFS.git
    
  2. Navigate to the project folder:

    cd NovaFS
    
  3. (Optional) Create and activate a virtual environment:

    Windows:

    python -m venv .venv
    .venv\Scripts\activate
    

    macOS/Linux:

    python3 -m venv .venv
    source .venv/bin/activate
    
  4. Install dependencies:

    (Note: this project uses only standard Python libraries.)


💻 Usage

1. Initialize the demo (first run only)

This creates the symbols/ files and a fresh symbol_map.json:

python init_demo.py

2. Launch the symbolic terminal

python nova_terminal.py

You'll get an interactive prompt. Enter a symbolic cue — a word like alpha, beta, growth or a glyph like 🔺, 🔵 — and NovaFS routes it to the associated memory file:

🔍 Enter a symbolic cue (e.g. 🔺, alpha, vision), or type 'exit': alpha

✅ FILE FOUND: 🔺_alpha.txt
🔗 ROUTE USED: alpha → 🔺_alpha.txt

Type exit to quit. Every routing event is logged for traceability.

3. Review the routing history

python view_trace.py          # human-readable memory trace
python route_log_viewer.py    # detailed route log

🗂️ File Descriptions

  • routing_engine.py: Core logic for cue resolution and logging
  • nova_terminal.py: Command-line interface
  • cue_map.json: Symbolic cue definitions
  • symbol_map.json: Mapping from symbols to file paths
  • route_log.json: Routing event log

📄 Example cue_map.json

{
  "alpha": {"symbol": "🔺", "weight": 1.0},
  "beta": {"symbol": "🔵", "weight": 1.0}
}

📄 Example symbol_map.json

{
  "🔺": {
    "paths": {
      "🔺_alpha.txt": 1
    }
  }
}

🧹 Resetting Logs

To clear all routing history:

  1. Open route_log.json
  2. Replace its contents with:
[]

🛡️ License & Disclaimer

Released under the MIT License.

This project is provided for demonstration purposes only.
No warranty is expressed or implied.


🌟 Vision

NovaFS demonstrates that even a minimal Python prototype can create a symbolic routing engine.
It offers a framework for future development in semantic search, AI memory systems, and adaptive context management.


About

NovaFS Symbolic Routing Demo

Resources

License

Stars

0 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages