Ticket-system

AI-Powered Home Appliance Query System

Overview

This project automates the processing, classification, and resolution of home appliance issues using AI. The system:

• Summarizes user queries and AI responses using Facebook BART.
• Classifies intents (troubleshooting, general inquiry, repair request).
• Stores classified intents in an SQLite database.
• Queries the database to fetch or log requests.
• Schedules an email for maintenance requests.
• Pipelines the entire workflow into a seamless process.

---

Workflow Pipeline

The project follows an end-to-end pipeline, with the following steps:

1. Convert JSON to CSV – Extract structured data from raw inputs.
2. Summarize Queries – Use Facebook BART to generate concise summaries.
3. Classify Intents – Assign intent types (troubleshooting, inquiry, repair).
4. Store Data in SQLite Database – Organize troubleshooting info, inquiries, and repair requests.
5. Query Database – Determine the next action based on intent.
6. Schedule Maintenance Emails – Add repair requests to an email queue.
7. Trigger AI Agent for Notifications – Process scheduled emails for task completion.

---

Project Structure

📁 data/                             # Data directory
   ├── home_appliance_query_response.csv  # Raw queries & responses
   ├── home_appliance_summary.csv         # Summarized data
   ├── home_appliance_summary_labeled.csv # Intent-classified data

📁 database/                         # SQLite database
   ├── home_appliance.db              # Stores troubleshooting, inquiries, repairs

📁 scripts/                           # All processing scripts
   ├── summarize_home_appliance.py    # Summarization using BART
   ├── qdslite.py                     # Intent classification & database storage
   ├── query_database.py              # Database interaction (fetch/log requests)
   ├── SE.py                          # Schedules maintenance request emails
   ├── test.py                        # AI Agent processing scheduled emails
   ├── run_pipeline.py                # End-to-end workflow execution

📄 README.md                          # Documentation

---

Setup & Installation

1. Install Dependencies

Ensure you have Python 3.8+ and install required packages:

pip install pandas transformers sqlite3

2. Set Up the Database

Initialize the SQLite database and required tables:

python query_database.py

3. Run the Complete Pipeline

Execute the entire AI pipeline in one command:

python run_pipeline.py

---

Workflow Details

Step 1: Summarization

• Converts JSON to CSV.
• Summarizes user-AI interactions using Facebook BART.
• Saves the output to home_appliance_summary.csv.

Step 2: Intent Classification & Database Storage

• Classifies intent types (troubleshooting, inquiry, repair).
• Stores classified data in SQLite database.

Step 3: Query Database for Next Steps

• Checks for troubleshooting & inquiries.
• Logs repair requests.

Step 4: Email Scheduling

• Adds pending repair requests to an email queue.

Step 5: AI Agent Processes Emails

• Retrieves scheduled emails.
• Sends or logs emails for maintenance confirmation.

---

Viewing Stored Data

Check All Troubleshooting Records

SELECT * FROM troubleshooting;

Check All Repair Requests

SELECT * FROM repair_requests;

Check Scheduled Emails

SELECT * FROM email_queue;

---

Features & Future Enhancements

✅ End-to-end AI pipeline for query processing & classification
✅ Automated database interactions via Text-to-SQL
✅ Email scheduling for repair requests
✅ Completely modular & scalable

Workflow Pipeline

Below is the workflow diagram for the project