TWC: TTP Option: Medium: Scalable Web Transparency: New Scientific Building Blocks, Tools, and Measurements to Tame the Data-Driven Web

随着“大数据”的出现，网络已经成为一个高度强大和不透明的地方，如果不加以控制，可能会滋生数据滥用和对个人的不公平做法。如今的Web服务积累了数量惊人的个人数据，如电子邮件、搜索和活动日志。它们是为了商业价值而挖掘的，在很长一段时间内保留，有时与第三方共享--所有这些都是在个人不知情或明确同意的情况下进行的。这些用途对个人来说是好是坏？它们是否与服务的隐私政策相匹配？数据删除后使用多长时间？谁与谁共享数据？目前，对这些问题的洞察还很少。该项目创建了一组构建块来跟踪不同类型的个人数据(例如，搜索中使用的关键字、以前访问的网站)，并衡量多种类型的使用(例如，定向广告)。我们的项目旨在通过开发工具和科学基础来提高数据驱动型Web的透明度，从而跟踪数据在Web上的旅程。我们在三个方面做出了贡献。首先，我们设计了一套构建块、高度可重复使用和可扩展的组件，这些组件有助于构建新一代审计工具，以揭开个人数据被如何使用的帷幕。其次，我们构建了一组健壮且可伸缩的透明度工具，这些工具实例化了这些构建块，并使用户、记者和调查人员能够了解Web服务的数据使用情况。我们正在通过与记者和调查人员的合作积极寻找部署。第三，我们利用这些工具对各种数据驱动平台进行广泛的测量研究，如定向广告生态系统、数据经纪人和在线价格歧视。这些研究提高了人们的意识，并帮助揭露了滥用数据的例子，我们希望这将为明智的社会就需要增加网络自愿透明度进行辩论提供依据。

With the advent of "big data," the Web has become a highly potent and opaque place which, left unchecked, can breed data misuse and unfair practices toward individuals. Today's Web services accumulate staggering amounts of personal data, such as emails, searches, and activity logs. They are mined for business value, retained for extended periods of time, and at times shared with third parties -- all without individuals' knowledge or explicit consent. Are these uses good or bad for individuals? Do they match a service's privacy policy? How long is data used after deletion? Who shares data with whom? Presently, there is little insight into these questions. This project creates a set of building blocks to track different types of individual data (e.g., keywords used in searches, previous sites visited) and measure multiple types of use (e.g., targeted advertising). These are used to develop robust and scalable tools for tracking data to uncover misuse.Our project seeks to increase the data-driven Web's transparency by developing both the tools and the scientific foundations for tracking the data's journey on the web. We make contributions on three fronts. First, we design a set of building blocks, highly reusable and scalable components that facilitate the building of a new generation of auditing tools to lift the curtain on how personal data is being used. Second, we build a set of robust and scalable transparency tools that instantiate those building blocks and enable users, journalists, and investigators to obtain visibility into Web services' data uses. We are actively seeking out deployments through collaborations with journalists and investigators. Third, we leverage these tools to run extensive measurement studies of various data-driven platforms, such as targeted advertising ecosystems, data brokers, and online price discrimination. These studies increase awareness, and help uncover examples of data mistreatment, which we hope will provide the grounds for an informed societal argument on the need for increased voluntary transparency on the Web.