|
|
--- |
|
|
title: Email Intent Classifier |
|
|
emoji: 🌍 |
|
|
colorFrom: red |
|
|
colorTo: indigo |
|
|
sdk: gradio |
|
|
sdk_version: 5.21.0 |
|
|
app_file: app.py |
|
|
pinned: false |
|
|
short_description: NLP model automatically classifying emails based on intent |
|
|
--- |
|
|
|
|
|
Email Intent Classifier: This application helps businesses manage email overload by automatically categorizing emails based on their intent. |
|
|
Problem: Email overload is a significant problem in businesses. Manually sorting and prioritizing emails takes valuable time from employees. Automatically categorizing emails by intent helps with: |
|
|
|
|
|
- Prioritization of urgent matters |
|
|
- Routing to appropriate departments |
|
|
- Providing quick responses |
|
|
- Managing workflow more efficiently |
|
|
|
|
|
Solution: This app uses a fine-tuned DeBERTa transformer model to classify emails into different intent categories: |
|
|
|
|
|
Question: Emails asking for information |
|
|
Request: Emails asking for action or approval |
|
|
Information: Emails sharing knowledge or updates |
|
|
Scheduling: Emails about setting up meetings or events |
|
|
Other: Any other intent |
|
|
|
|
|
How to Use: |
|
|
- Enter or paste an email text into the input box with subject |
|
|
- Click "Submit" |
|
|
- View the predicted intent categories and their probability/confidence scores |
|
|
|
|
|
Model Details: This model was fine-tuned on an email dataset with the following specifications: |
|
|
|
|
|
Base Model: DistilBERT (base-uncased) |
|
|
Training Dataset: Synthetic dataset of categorized emails across 6 intent categories |
|
|
|
|
|
Limitations: |
|
|
Very short or ambiguous emails may be classified with lower confidence |
|
|
Emails with multiple intents may be classified based on the dominant intent |
|
|
|
|
|
Deployment: |
|
|
This application is deployed using Hugging Face Spaces with Gradio. The model weights are stored in the Hugging Face Model Hub. |
|
|
|
