Mobile app user reviews contain valuable feedback about software functionality, quality, and security. Existing approaches for analyzing security-related reviews often rely on traditional feature extraction methods, limiting their ability to detect nuanced security concerns. This study aims to develop an automated framework for classifying mobile app user reviews into security-related and non-security-related categories.
This repository introduces PPMC (Perturbed Principal Components Mechanism), a tunable anonymization method based on Principal Component Analysis (PCA), designed to support privacy-preserving analytics in Internet of Medical Things (IoMT) environments. The proposed approach employs controlled eigenvector perturbation to achieve a principled trade-off between data utility and privacy protection, without injecting noise directly into the raw feature space.
The ComFASE attack injection campaign dataset contains summarized results from a large-scale jamming attack testing campaign on connected automated vehicle platoons. Each row corresponds to one experiment and includes the attack configuration (experiment ID, attack start and end times, injected jamming value), the observed impact (impact status, collision occurrence, collision time, and colliding vehicle), and vehicle-level dynamics captured as maximum acceleration and minimum deceleration values for each vehicle in the platoon.
This dataset contains anonymized raw data collected during the design, implementation, and evaluation of The XSS Game (TXG), a web-based educational game developed to support learning of cross-site scripting (XSS) vulnerabilities through multi-opposing-role gameplay, narrative-based simulations, and structured in-game feedback. The dataset covers two iterative versions of the game (TXG V1.1 and TXG V2.1) evaluated as part of a Design-Based Research study.
The dataset consists of well-organized and curated source code samples that are meant for automated detection, classification, and fixing up of software security vulnerabilities. It has marked occurrences of both AI-generated and human-written Python code showing up with the same security flaws like - SQL Injection, Command Injection, Cross-Site Scripting (XSS), Path Traversal, Hardcoded Secrets, and Unsafe Dynamic Code Execution (eval/exec).
The PGA dataset is a safety alignment dataset specifically designed for Large Reasoning Models (LRMs). It contains 3,989 rigorously verified high-quality samples, aimed at addressing the unique Reasoning-Activated Jailbreak problem in LRMs. Unlike traditional Simple Refusal datasets, PGA strives to shift the alignment paradigm toward Intrinsic Safe Reasoning, thereby enhancing model safety while preserving deep reasoning capabilities.
BJTU-RAO 转向架数据集由北京交通大学先进轨道自主运行全国重点实验室面向全球学者、工程师及研究生公开发布，系全球首个轨道交通列车转向架传动系统故障模拟公开数据集。该数据集通过地铁列车转向架传动系统故障模拟实验获得，包含 51 种健康状态的多传感器数据。BJTU-RAO 转向架数据集是完全公开的，任何研究人员皆可以通过后文的下载链接免费下载数据，且在标注数据来源的前提下，研究人员可以自由使用 BJTU-RAO 转向架数据集。例如，使用该数据集验证各类电机、齿轮箱和轴箱的故障检测及诊断算法。
This dataset provides a curated collection of stage-wise rug pull attack sequences extracted from real-world DeFi incidents.
Each record represents a complete attack trajectory, explicitly linking a preparation-stage transaction to its corresponding execution-stage transaction through a common attacker-controlled address.
The UIDS-CAN dataset is a multi-vehicle Controller Area Network (CAN) dataset designed for universal intrusion detection system (UIDS) research. It contains real-world CAN traffic collected from heterogeneous vehicle platforms under both normal driving conditions and multiple cyber-attack scenarios.
Vehicular Named Data Networking (VNDN) has emerged as a promising solution to improve
content lookup, scalability, and reliability in dynamic vehicular environments. However, the name-centric
communication model and reliance on cooperative forwarding introduce new security challenges, making
VNDN vulnerable to different attacks, such as forwarding-layer attacks that cannot be fully addressed
through traditional cryptographic mechanisms. In the meantime, there is still a lack of publicly available
