sattipoe/AI-Assisted-Aspect-Based-Sentiment-Classification-System

LuxeHaven ABSA Lab (Aspect-Based Sentiment Analysis)

This Space runs the Week 8 ABSA notebook as a production Gradio app.

Input CSV schema

Your CSV must include:

• —Review (text)
• —Aspects (gold labels like {Ambience: Positive, Service: Negative, Food: Positive})

Setup

1. 1.Add a Space secret:
2. 2.OPENAI_API_KEY = your OpenAI API key
3. 3.Upload the CSV and click Run ABSA

What you get

• —Predictions table with a JSON predicted_aspects column
• —Downloadable CSV of predictions
• —Proper ABSA metrics:
• —Aspect detection precision/recall/F1
• —Sentiment macro/micro F1
• —Exact match rate
• —Per-aspect accuracy/F1
• —Confusion matrix labels (includes missing when the model doesn't output an aspect)

Cost control

Use Max rows to limit how many reviews are sent to the model.

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference