Claude Data Cleaner

A Flask web application for cleaning and standardizing company names and addresses using Claude AI.

Features

• Upload CSV files containing company names and addresses
• Clean and standardize data using Claude AI
• Support for multiple Claude models (Haiku and Sonnet)
• Preserve reference columns
• Download cleaned results and error reports
• Real-time processing status and logs

Requirements

• Python 3.12+
• Anthropic API key

Usage

1. Upload a CSV file containing at minimum:
   • name column: Entity names
   • address column: Full addresses
2. Optional columns:
   • country: Country names/codes
   • Any additional reference columns
3. Select Claude model:
   • Haiku (faster)
   • Sonnet (more thorough)
4. Monitor processing progress
5. Download results when complete

File Format

Input CSV Format

name,address,country,id
Acme Corp,123 Main St,US,001

Output CSV Format

The tool will generate a CSV with the following columns:

• All original columns
• cleaned_name_1: Primary standardized name
• cleaned_name_2: Alternative name (if DBA/AKA present)
• cleaned_address: Standardized address
• cleaned_country: Standardized country
• has_dba: Multiple names indicator
• country_match: Indicator if country match
• name_inferred: Indicator if null name inferred
• status: Processing status
• error: Error message (if any)

Limitations

• Maximum file size: 16MB
• Supported file format: CSV only
• Required columns: name, address