Skip to content
プロジェクト

ali-faraz-py/SentimentSense has 26 GitHub stars and 0 forks. GitHot measured 0 stars in 24 hours and 0 in 7 days. Last checked 2026-11-21 20:29 UTC. Last push 2026-07-25.

この数値の測り方

できること

  • Dual-Model Pipeline: Combines a DistilBERT transformer for sentiment with a BART-Large model for zero-shot topic classification.
  • Aspect-Based Analysis (ABSA): Automatically identifies specific categories (Food, Service, Price, etc.) mentioned in the text without manual labeling.
  • Contextual Nuance: Correctively handles negations (e.g., "not great") and complex linguistic structures that traditional models miss.
  • Real-Time Inference: High-speed processing using optimized PyTorch backends.

計測したスター増加

24時間
0
7日間
0
15日間
履歴を収集中
1か月
履歴を収集中

期間ごとの純増数です。期間は重複しており、日別の値ではありません。未計測の値は空欄です。スター数は注目度の指標であり、品質の保証ではありません。

概要

🧠 Sentiment Sense: Advanced NLP dashboard using DistilBERT & Zero-Shot (BART) for Aspect-Based Sentiment Analysis. Detects business topics (Food, Service, Price) & emotional context in real-time. Built for enterprise feedback processing. 🚀

機械学習

🧠 Sentiment Sense: Advanced Aspect-Based Sentiment AI

A professional-grade Natural Language Processing (NLP) dashboard built with Python and Streamlit. This tool utilizes Transfer Learning via Hugging Face Transformers to perform simultaneous sentiment analysis and aspect extraction with high contextual awareness.


🚀 Live Demo

Click here to try the Live App


📺 Demo Preview

Sentiment Sense Demo


✨ Features

  • Dual-Model Pipeline: Combines a DistilBERT transformer for sentiment with a BART-Large model for zero-shot topic classification.
  • Aspect-Based Analysis (ABSA): Automatically identifies specific categories (Food, Service, Price, etc.) mentioned in the text without manual labeling.
  • Contextual Nuance: Correctively handles negations (e.g., "not great") and complex linguistic structures that traditional models miss.
  • Real-Time Inference: High-speed processing using optimized PyTorch backends.
  • Enterprise UI: Interactive metrics, progress bars, and sidebar technical documentation for an intuitive user experience.

🛠️ Tech Stack

  • Language: Python 3.12+
  • Framework: Streamlit (Web UI)
  • NLP Engines: Hugging Face Transformers
  • Deep Learning Framework: PyTorch
  • Models: distilbert-base-uncased & facebook/bart-large-mnli
  • Deployment: Streamlit Community Cloud

🚀 Installation & Local Setup

  1. Clone the repository:

    git clone [https://github.com/ali-faraz-py/SentimentSense](https://github.com/ali-faraz-py/SentimentSense)
    cd SentimentSense
    
  2. Install dependencies:

     pip install -r requirements.txt
    
  3. Run the application:

     streamlit run app.py
    

📂 Project Structure

sentimentsense/
├── app.py              # Main Streamlit application and dual-model logic
├── requirements.txt    # Project dependencies (Transformers, Torch, Streamlit)
├── .gitignore          # Prevents tracking of cache and environment files
└── explore.ipynb       # Initial research on LSTM-based sentiment analysis

🧠 Model Insights

The engine moves beyond simple "keyword matching" by using Self-Attention mechanisms.

  • The Sentiment Stage: Uses DistilBERT, which was trained on the SST-2 dataset, allowing it to achieve 90%+ accuracy on general sentiment tasks while remaining lightweight.
  • The Aspect Stage: Uses Zero-Shot Classification (BART). This allows the app to detect any custom category (like "Atmosphere" or "Health") by measuring the semantic similarity between the input text and the target labels.

👤 Author

Syed Ali Faraz - GitHub Profile

If you found this NLP pipeline useful, please give the repository a ⭐!

関連プロジェクトを探す

読み込み中…