UberJugaad Enhanced SALT Dataset visualization

Dataset Overview

The UberJugaad Enhanced SALT Dataset transforms SAP's transactional SALT dataset into a complete enterprise simulation by layering realistic business communications, documents, and embedded patterns. This comprehensive dataset provides researchers with a realistic business ecosystem for developing and testing AI/ML applications.

DOI Citation Kaggle Dataset GitHub Repository

Dataset Statistics

1.9M

ERP Transactions

151K

Business Emails

3.5K

Supporting Documents

159MB

Total Dataset Size

About UberJugaad GmbH

UberJugaad GmbH is a fictional EUR 14.8B German industrial supplier created for this dataset. The company represents a realistic enterprise environment with complex business operations, customer relationships, and communication patterns.

€14.8B

Annual Revenue

1,467

Employees

15,606

Customers

2019-2020

Data Period

Key Features

🔗 Linked Data

All files connected via SALESDOCUMENT and customer_id for comprehensive analysis

📧 Realistic Communications

151K emails with authentic subjects, bodies, and business context

⏰ Time-Synchronized

Chronologically consistent data spanning 2019-2020 business operations

🔍 Discovery-Oriented

Patterns embedded in content rather than explicit labels for natural learning

💼 Complete Business Ecosystem

Email threads, documents, and transactions forming realistic business scenarios

🗃️ Multiple Formats

Parquet files for analysis, SQLite database for queries, comprehensive documentation

Dataset Structure

The dataset is organized into logical groups for easy access and analysis:

📊 Core Data Files (83.5 MB)

all_communications.parquet - 151,673 business emails (6.9 MB)
erp_transactions.parquet - 1.9M ERP transactions (40.8 MB)
sales_documents.parquet - 243K sales headers (7.1 MB)
sales_items.parquet - 1.9M line items (28.4 MB)
supporting_documents.parquet - 3,467 supporting documents (0.1 MB)
business_documents.parquet - 32 business reports (22 KB)

📧 Email Database (76 MB)

uberjugaad_email.db - SQLite database with 151K emails + 19K contacts

📖 Documentation & Code

Comprehensive guides, column descriptions, and getting started materials
Sample Jupyter notebook for data exploration
Email examples and business document conversion guides

Use Cases

Email Classification & Routing

Sentiment Analysis

Customer Behavior Analysis

Document Information Extraction

Business Process Mining

Anomaly Detection

Natural Language Understanding

Multi-modal Analysis

Quick Start

Get started with the dataset using this simple Python code:

import pandas as pd

# Load emails
emails = pd.read_parquet('all_communications.parquet')
print(f"Total emails: {len(emails):,}")

# Load transactions  
transactions = pd.read_parquet('erp_transactions.parquet')
print(f"Total orders: {len(transactions):,}")

# Find emails mentioning specific orders
order = "0002315309"
order_emails = emails[emails['body'].str.contains(order, na=False)]
print(f"Emails about order {order}: {len(order_emails)}")

🚀 Enhanced for AI Research

This dataset builds on SAP's SALT dataset with synthetic but realistic business communications modeled on real-world enterprise patterns. Perfect for developing and testing AI applications in realistic business contexts.

Citation

If you use this dataset in your research, please cite:

@dataset{rutledge_patrick_2025_17125322,
  author       = {Rutledge, Patrick},
  title        = {{UberJugaad GmbH Enhanced SALT Dataset}},
  month        = sep,
  year         = 2025,
  publisher    = {Zenodo},
  version      = {v1.0.0},
  doi          = {10.5281/zenodo.17125322},
  url          = {https://doi.org/10.5281/zenodo.17125322}
}

APA Format:
Rutledge, P. (2025). UberJugaad GmbH Enhanced SALT Dataset (Version v1.0.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.17125322

License & Acknowledgments

License: MIT License for enhanced content / SAP Sample Code License for original SALT data

Acknowledgments: Built on SAP's SALT dataset, enhanced with synthetic business communications for AI/ML applications. The original SALT authors are not affiliated with these enhancements.

View DOI Citation Access on Kaggle Source Code