Enron Email Dataset

Overview

• Source: The dataset originates from 600,000 emails released by the U.S. Federal Energy Regulatory Commission, involving 158 employees. It was later purchased and processed by MIT, with some attachments deleted or edited.
• Versions: Versions of the dataset are available at the Library of Congress and https://www.cs.cmu.edu/~./enron/.

Subset

• Subset Source: Multiple subsets of the dataset can be found online, including on GitHub, HuggingFace, and Kaggle.
• Specific Subset: Researchers from the Institute of Informatics and Telecommunications of Greece described a commonly used subset in their paper [Metsis]. This subset selected six Enron employees with large email volumes, containing 33,716 emails, of which 17,171 are spam.

Data Content

• File Format: CSV file.
• Fields: Includes a concatenated subject‑and‑body field and a separate original filename column.

Research Purpose

• Research Direction: Used to analyze and test various spam filters, including multiple Naïve Bayes versions.

References

• [Metsis] Metsis, V., Androutsopoulos, I., & Paliouras, G. "Spam filtering with naive bayes‑which naive bayes?" Proceedings of the 3rd Conference on Email and Anti‑Spam (CEAS 2006), Mountain View, CA, USA, 2006.