URL Phishing Detector

Model Description

Random Forest ๋ชจ๋ธ๋กœ URL์˜ ์•…์„ฑ ์—ฌ๋ถ€๋ฅผ ํŒ๋‹จํ•˜๋Š” ์ด์ง„ ๋ถ„๋ฅ˜๊ธฐ์ž…๋‹ˆ๋‹ค.

ํ”„๋กœ์ ํŠธ: ์Šค๋ฏธ์‹ฑ ํƒ์ง€ ์‹œ์Šคํ…œ (OCR/QR + URL ์œ„ํ—˜ ํŒ๋‹จ)
์ž‘์„ฑ์ž: ์ด๊ฒฝ์ค€

Performance

  • Accuracy: 100.00%
  • Precision: 100.00%
  • Recall: 100.00%
  • F1-Score: 100.00%

Features

30๊ฐœ์˜ URL ํŠน์ง•์„ ์‚ฌ์šฉํ•ฉ๋‹ˆ๋‹ค:

  • URL ๊ตฌ์กฐ (๊ธธ์ด, ํŠน์ˆ˜๋ฌธ์ž, ํ•˜์ดํ”ˆ ๋“ฑ)
  • ๋„๋ฉ”์ธ ํŠน์„ฑ (IP ์ฃผ์†Œ, ์—”ํŠธ๋กœํ”ผ, TLD ๋“ฑ)
  • ์ฝ˜ํ…์ธ  ํŠน์ง• (HTTPS, ์˜์‹ฌ ํ‚ค์›Œ๋“œ, ๋ธŒ๋žœ๋“œ ๋ถˆ์ผ์น˜ ๋“ฑ)

Usage

import joblib
from huggingface_hub import hf_hub_download

# ๋ชจ๋ธ ๋‹ค์šด๋กœ๋“œ
model_path = hf_hub_download(repo_id="kanell0304/url-phishing-detector", filename="url_classifier.pkl")
model = joblib.load(model_path)

# ์˜ˆ์ธก
# features = extract_features(url)  # ํŠน์ง• ์ถ”์ถœ ํ•„์š”
# prediction = model.predict([features])

Training Data

  • Phishing URLs: PhishTank, URLhaus
  • Benign URLs: Tranco Top Sites

Limitations

  • ์ƒˆ๋กœ์šด ํ”ผ์‹ฑ ํŒจํ„ด์— ๋Œ€ํ•œ ์ง€์†์ ์ธ ์žฌํ•™์Šต ํ•„์š”
  • ๋‹จ์ถ• URL์€ ํ™•์žฅ ํ›„ ๋ถ„์„ ๊ถŒ์žฅ

Citation

@misc{url-phishing-detector,
  author = {์ด๊ฒฝ์ค€},
  title = {URL Phishing Detector},
  year = {2025},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/kanell0304/url-phishing-detector}}
}
Downloads last month
-
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support