{
 "id": "NWeak/emails-user-study",
 "owner": "NWeak",
 "name": "emails-user-study",
 "downloads": 29,
 "likes": 0,
 "last_modified": "2026-09-22T12:22:59.000Z",
 "stated_licence": "cc-by-4.0",
 "stated_from": "tag",
 "licence_name": null,
 "licence_link": null,
 "file_licence": null,
 "licence_files": [],
 "sources": [],
 "gated": false,
 "tasks": [],
 "citation": "The emails themselves are **not ours**. They come from the\n**PhishingSpamDataSet** of Toth, Bisztray and Dubniczky, released under\nCC-BY-4.0 and redistributed here under that licence:\n\n> Rebeka Toth, Tamas Bisztray, Richard A. Dubniczky.\n> *Constructing and Benchmarking: a Labeled Email Dataset for Text-Based\n> Phishing and Spam Detection Framework.*\n> [arXiv:2511.21448](https://arxiv.org/abs/2511.21448) |\n> [github.com/DataPhish/PhishingSpamDataSet](https://github.com/DataPhish/PhishingSpamDataSet)\n\nTheirs: `Subject`, `Body`, `Sender`, `URL(s)`, `Type`, `Created by`, `Source`,\n`Year`, and the `LLM detected emotion` / `LLM detected motivation`\nannotations. Ours: the filtering and split, the embeddings, the 6-concept\nmerge, and the trained models' outputs.\n\nPlease cite Toth et al. alongside our paper if you use this data.\n\n**Note:** `Sender` and `Body` are real email content carried over from the\nsource dataset.\n\n\nProduced for **[Are Concept Bottleneck Models Effective as Decision-Support\nSystems?](https://arxiv.org/abs/2608.25581)** (arXiv:2608.25581) -- Bogani,\nDebole, Marconato, Pugnana, Tentori, Passerini. Code:\n[github.com/debryu/user-study-CBMs](https://github.com/debryu/user-study-CBMs).",
 "signal": "permissive",
 "reason": "The declared licence (cc-by-4.0) is classified as permissive.",
 "revision": "76b11d89a872c903bc699df8f8da82d55c4c28a8",
 "stated_evidence": "https://huggingface.co/datasets/NWeak/emails-user-study/blob/76b11d89a872c903bc699df8f8da82d55c4c28a8/README.md",
 "file_evidence": "https://huggingface.co/datasets/NWeak/emails-user-study/tree/76b11d89a872c903bc699df8f8da82d55c4c28a8",
 "card_nc": false
}