kms-engineer/assistant-bot-intent-classifier

Intent Classifier for Contact Management Assistant Bot

This model is a fine-tuned RoBERTa-base model for intent classification in contact management tasks.

Model Description

• Developed by: Mykyta Kotenko
• Base Model: roberta-base by Facebook AI
• Task: Text Classification (Intent Recognition)
• Language: English
• License: MIT

Supported Intents

This model recognizes 15+ different intents for contact management:

Contact Management

• add_contact - Add new contact with name, phone, email, address, birthday
• edit_phone - Update contact's phone number
• edit_email - Update contact's email address
• edit_address - Update contact's address
• delete_contact - Delete a contact
• show_contact - Show details of a specific contact
• show_contacts - List all contacts
• search_contacts - Search for contacts

Notes

• add_note - Add a note to a contact
• show_notes - Show all notes or notes for a contact
• edit_note - Edit an existing note
• delete_note - Delete a note

Tags

• add_tag - Add a tag to a contact
• remove_tag - Remove a tag from a contact

Other

• show_birthdays - Show upcoming birthdays
• help - Show help message
• exit - Exit the application

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline

# Load model and tokenizer
tokenizer = AutoTokenizer.from_pretrained("kms-engineer/assistant-bot-intent-classifier")
model = AutoModelForSequenceClassification.from_pretrained("kms-engineer/assistant-bot-intent-classifier")

# Create classification pipeline
classifier = pipeline("text-classification", model=model, tokenizer=tokenizer)

# Classify intent
text = "add contact John Smith 212-555-0123 john@example.com"
result = classifier(text)
print(result)
# Output: [{'label': 'add_contact', 'score': 0.98}]

# More examples
examples = [
    "update phone for Sarah to 555-1234",
    "show all my contacts",
    "delete contact Bob",
    "add note for Alice: Call back tomorrow"
]

for text in examples:
    result = classifier(text)
    print(f"{text} → {result[0]['label']} ({result[0]['score']:.2f})")

Training Details

• Base Model: roberta-base
• Training Dataset: Custom dataset with contact management commands
• Learning Rate: 2e-5
• Batch Size: 16
• Epochs: 3-5
• Optimizer: AdamW

Intended Use

This model is designed for:

• Contact management applications
• Personal assistant bots
• CRM systems with natural language interface
• Voice-controlled contact management

Limitations

• Optimized for English language only
• Best performance on contact management domain
• May not generalize well to other domains without fine-tuning

Example Predictions

Input: "add new contact John Doe 555-1234 john@email.com"
Output: add_contact (confidence: 0.99)

Input: "change email for Sarah to sarah@newmail.com"
Output: edit_email (confidence: 0.97)

Input: "show me all contacts"
Output: show_contacts (confidence: 0.98)

Input: "delete contact Bob"
Output: delete_contact (confidence: 0.96)

Input: "add tag 'work' to Alice"
Output: add_tag (confidence: 0.95)

Model Architecture

Based on RoBERTa (Robustly Optimized BERT Pretraining Approach):

• 12 transformer layers
• 768 hidden dimensions
• 12 attention heads
• ~125M parameters

Citation

If you use this model, please cite:

@misc{kotenko2025intentclassifier,
  author = {Kotenko, Mykyta},
  title = {Intent Classifier for Contact Management Assistant Bot},
  year = {2025},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/kms-engineer/assistant-bot-intent-classifier}},
  note = {Based on RoBERTa by Facebook AI}
}

Acknowledgments

• Base Model: RoBERTa by Facebook AI Research
• Framework: Hugging Face Transformers
• Inspiration: Contact management and personal assistant applications

License

MIT License - See LICENSE file for details.

This model is a derivative work based on RoBERTa, which is licensed under MIT License by Facebook, Inc.

Contact

• Author: Mykyta Kotenko
• Repository: assistant-bot
• Issues: Please report issues on GitHub