SaTC: CORE: Small: Securing Web-to-Mobile Interface Through Characterization and Detection of Malicious Deep Links

随着移动设备的广泛采用，移动网站和应用程序(App)已成为访问在线内容的主要界面。移动深度链接是索引移动内容的关键机制，已被广泛用于使用户能够在网站、搜索引擎和移动应用程序之间导航，并成为改善用户在线体验的工具。不幸的是，移动Deeplink在设计时没有考虑到安全性，它允许恶意应用程序或网站发起秘密攻击，以劫持用户点击、窃取敏感信息和执行网络钓鱼攻击。随着每天都有更多的应用程序和网站深度链接，这对广大互联网人口来说是一个新的安全威胁。该项目正在开发方法和可用的工具来测量、分析和保护当今移动网络生态系统中的移动深度链接。为了首先了解新出现的安全威胁，研究人员正在开发新的测量框架，通过自动从市场规模的应用程序中提取移动深度链接来绘制移动应用程序和网络之间的经验联系。此外，研究人员正在研究自动化技术，以检测基于深度链接的滥用行为，而不需要依赖大规模的地面事实数据。一个目标方法是为应用程序和网站构建深度链接图，并使用半监督学习和信任比例分配方案探索检测技术。最后，研究人员正在开发分析工具，从移动深度链接中提取上下文信息，以使智能警报系统能够保护深度链接的使用。研究人员正在研究安全性和可用性之间的权衡，以设计在不牺牲可用性的情况下使用Deeplink的更安全的方式。

With the wide adoption of mobile devices, mobile websites and applications (apps) have become the primary interfaces to access online content. Mobile deep-linking, a key mechanism to index mobile content, has been widely used to enable users to navigate across websites, search engines and mobile apps and become instrumental to improving users' online experience. Unfortunately, mobile deeplinks are designed without security in mind, which allows malicious apps or websites to launch stealthy attacks to hijack user clicks, steal sensitive information and perform phishing attacks. With more apps and websites deep-linked every day, this becomes an emerging security threat to a broad Internet population. This project is developing methodologies and usable tools to measure, analyze and secure mobile deeplinks in today's mobile-web ecosystem.To first understand the emerging security threat, the researchers are developing novel measurement frameworks to map out the empirical connections between mobile apps and the web by automatically extracting mobile deeplinks from the market-scale apps. In addition, the researchers are investigating automated techniques to detect deeplink-based abuse without relying on large-scale ground-truth data. One target approach is to construct deeplink graphs for apps and websites and explore detection techniques using semi-supervised learning and trust prorogation schemes. Finally, the researchers are developing analytics tools to extract context information from mobile deeplinks to enable intelligent alert systems to safeguard the deeplink usage. The researchers are investigating the tradeoffs between security and usability to design safer ways of using deeplinks without sacrificing the usability.