Lost in the Digital Abyss: How I Created a Tool to Rescue My Files

LostFile_doc_2086334104.doc
LostFile_doc_2086334105.doc
LostFile_doc_2088324242.doc
LostFile_docx_2091254245.docx
… (x10000 👀)

We’ve all been there: staring at a screen full of cryptically named files after an unexpected data loss. This weekend, I found myself in that exact nightmare, courtesy of Windows Media Creation Tool (thanks a lot 😒).

Here’s the thing about file recovery: even when files are “deleted,” the data often remains. The catch? You lose the original file structure and names. Frustrating, right?

So, I did what any tech-savvy problem solver would do — I built a tool.

🛠️ What Makes This Tool Special?

• Uses AI (specifically Mistral 7B 0.3) to analyze documents
• Generates meaningful file names
• Keeps all your data local and private (no transfer on another private server or other)
• Recovers probable document dates

How It Works:

When you feed your lost files into the tool, it first uses an AI model (Mistral 7B 0.3) to perform a deep content analysis. This isn’t just a simple file scan — it’s an intelligent extraction process.

We will read the file content and ask the LLM two questions:

• Find this document’s date
• What would be a fitting title for this document

And that’s it ! We can turn a file like “LostFile_doc_2086334104.doc” into something readable like “2024–08–17 Resignation Letter.doc”

Proof that machine learning isn’t just hype — it’s a practical problem-solver!

📌 Check out the tool: https://gitlab.com/datatoy/textfilerenamer

#MachineLearning #DataRecovery #AI #Productivity