bert-it-issue / README.md
SalomonMetre13's picture
Update README.md
cec509d verified
metadata
license: mit
language:
  - en
metrics:
  - accuracy
base_model:
  - google-bert/bert-base-uncased
pipeline_tag: text-classification

BERT for IT Support Ticket Classification

A fine-tuned BERT model for classifying IT-related support tickets into predefined categories.

Model Details

  • Model Name: bert-it-issue
  • Author: SalomonMetre13
  • Architecture: BERT-based model fine-tuned for IT support ticket classification
  • Task: Text Classification (text-classification)
  • Dataset: Processed IT support tickets dataset (all_tickets_processed_improved_v3.csv)
  • Labels:
    • 0: Hardware
    • 1: Access
    • 2: Miscellaneous
    • 3: HR Support
    • 4: Purchase
    • 5: Administrative rights
    • 6: Storage
    • 7: Internal Project

Usage

You can use this model for automatically classifying IT support requests based on their content.

Example Usage with Transformers (Python)

from transformers import pipeline

classifier = pipeline("text-classification", model="SalomonMetre13/bert-it-issue")

text = "I need a new laptop because mine stopped working."
prediction = classifier(text)

print(prediction)  # [{'label': 'Hardware', 'score': 0.97}]

Using Hugging Face API (cURL)

curl -X POST "https://api-inference.huggingface.co/models/SalomonMetre13/bert-it-issue" \
     -H "Authorization: Bearer YOUR_HF_API_TOKEN" \
     -H "Content-Type: application/json" \
     -d '{"inputs": "I need access to my email account."}'

Performance

The model was trained and evaluated on a dataset of categorized IT support tickets, achieving high accuracy on validation and test sets.

Applications

  • Automated IT ticket classification
  • Helpdesk support systems
  • Chatbot integration for IT requests

Limitations

  • May misclassify ambiguous requests.
  • Performance depends on how well the training data represents real-world IT tickets.
  • Doesn't handle multi-label classification (only assigns one category per ticket).

Contributions & Feedback

Feel free to contribute by fine-tuning, reporting issues, or suggesting improvements! 🚀