Apppublic

Baphomert/ScentSational-Fragrantica-LFS2

sourceHugging Faceupdated 9mo agoView on Hugging Face

0likes

App README

✨ SCENTSATIONAL | AI CORE

The Intelligence Platform. Unlock the chemical DNA of scent.

![Open in Spaces](https://huggingface.co/spaces/Baphomert/ScentSational-Fragrantica-LFS2)   ![Visit Atelier Dashboard](https://github.com/MagdalenaRomaniecka/ScentSational)

</div>

💎 Project Overview

ScentSational AI Core is the backend intelligence engine designed to revolutionize fragrance discovery. Unlike traditional filters that rely solely on checkboxes, this engine uses Semantic Search and Natural Language Processing (NLP) to understand the vibe of a scent (e.g., "dark, woody notes with a luxury feel").

This repository serves as the computational brain, separating heavy logic from the visual presentation layer.

📐 Architecture & Ecosystem

This project is part of a dual-repository architecture designed for scalability and clean code separation.

PART A: THE FACE (Frontend)	PART B: THE BRAIN (Backend)
ScentSational Atelier	ScentSational AI Core (This Repo)
🎨 Interactive Streamlit Dashboard	🧠 Hugging Face Space & LFS Storage
Presentation Layer	Computation Layer

⚙️ Key Features & Tech Stack

This engine translates abstract concepts into mathematical vectors using state-of-the-art ML libraries.

—Sentence-Transformers (SBERT): Generates dense vector representations (embeddings) of perfume notes.
—Cosine Similarity: Calculates the mathematical distance between scent profiles to find the nearest neighbors.
—Git LFS (Large File Storage): Used to host heavy model artifacts required for inference.
—Stack: Python 3.9+, Streamlit, Pandas, Numpy, Scikit-Learn.

💻 Installation & Usage

To run this backend engine locally:

bash

# 1. Clone the repository (ensure Git LFS is installed)
git clone [https://github.com/MagdalenaRomaniecka/ScentSational-Fragrantica-LFS.git](https://github.com/MagdalenaRomaniecka/ScentSational-Fragrantica-LFS.git)

# 2. Install dependencies
pip install -r requirements.txt

# 3. Run the application
streamlit run app.py

---

📂 Data Source & LFS Structure

Due to GitHub's file size limits, this repository uses Git LFS to host artifacts.

File Name	Description	Role
`scent_embeddings.pkl`	Vectorized representation of 40k+ perfumes	Semantic Database
`scentsational_data.csv`	Raw metadata (Brand, Notes, Ratings)	Source Data

Data Source: This project utilizes the Fragrantica Perfumes Dataset sourced from Kaggle.

👩‍💻 Author

Created by Magdalena Romaniecka <br> Data Analyst & Web Analytics Enthusiast

<br>

</div>