"phishing 2026 Dataset"
Year: 2026
Type: dataset
Abstract: "The Proposed Multi-Domain Phishing URL Feature Dataset (PMPUFD) is a structured dataset designed for research on phishing URL detection, feature selection, classification, optimization, ablation analysis, and robustness evaluation. The validated dataset contains 911,728 URL records and 187 columns, with 'URL' representing the URL field and 'label' representing the classification target.The feature space is organized into multiple analytical domains, including URL structure, domain and subdomain characteristics, character and ratio statistics, entropy and randomness, protocol and TLD indicators, IP and port characteristics, obfuscation and encoding, webpage\/HTML characteristics, URL-domain-title similarity, brand\/similarity indicators, typosquatting and attack-pattern indicators, and homograph\/Unicode indicators where represented by the submitted columns.PMPUFD is intended to support systematic investigation of feature relevance, redundancy, feature selection, and phishing URL classification. In particular, the dataset provides an experimental foundation for evaluating Information Gain-guided Heuristic-Constrained Feature Selection (IG-HC\u00b2FS) and subsequent hyperparameter or multi-objective optimization strategies.The dataset should be used with a leakage-safe experimental protocol. For final predictive experiments, preprocessing parameters, Information Gain calculations, correlation thresholds, feature-selection decisions, and optimization decisions should be fitted using the training partition only. Validation data should be used for model-selection decisions, while the held-out test partition should remain unseen until final evaluation.The dataset is provided as a research resource for phishing URL detection experiments. Performance superiority should be established through controlled baseline comparisons, ablation studies, robustness testing, and appropriate statistical evaluation rather than inferred from the dataset composition alone. "
Source: IEEE DataPort
DOI: 10.21227/7zkt-a868
Author: ASHVINI JADHAV
Language: English
Topic: Spam and Phishing Detection
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7204654167