NetSE: Medium: Collaborative Research: Auditing Internet Content for Credibility, Fairness, and Privacy

每天，数以百万计的用户正在访问互联网上数十亿网页和其他内容中的一部分。虽然网络研究社区和公众一般都很清楚网络中立性问题，即如何制定监管政策和审计机制，以防止互联网服务提供商歧视各种应用程序，但在实现内容中立性方面投入的努力很少。特别是，我们的几乎每次浏览点击都被基于网络或互联网服务提供商的“信息收集和聚合者”收集，我们的个人资料被用于在线广告，这已经不是什么秘密了。然而，在这一新兴领域还没有能够发现并告知最终用户此类做法的公共审计机制。再以搜索引擎为例。人们如何知道这些服务上提供的信息不会因为商业、政治或任何其他原因而有偏见（或将来会有偏见）？更普遍的是，没有任何公共审计系统能够监测互联网上广告或垃圾邮件活动的范围和有效性。pi认为，所有这些问题都从根本上影响到各个层面的公平、信息可信度和用户隐私，所有这些都与互联网的未来发展及其他方面有着重要的关系。pi将在一个系统中建立一套统一的方法和工具，该系统能够(i)为网络广告领域启用审计机制，（ii）监控搜索引擎。服务并揭示其中立性，以及（iii）独立确定网站的受欢迎程度并检查所宣传的受欢迎程度的真实性。此外，pi计划将开发的系统部署在PlanetLab广域网测试平台上，以执行互联网内容中立性的长期监测。更广泛的影响：通过告知公众在线广告商和搜索引擎的行为，审计系统将促进公平竞争，并排除垄断、寡头垄断或共谋，最终有助于提高在线信息的可信度和用户隐私。即使在今天的互联网上不存在偏见或侵犯隐私的行为（不幸的是，PI的初步结果表明情况并非如此），审计系统和方法也将永远是一个障碍，因为有人认为应用这种方法可能不会引起公众的注意和反响。通过揭露有偏见的内容行为，审计系统将能够揭露可能因商业、社会、政治、宗教或任何其他原因而发生的歧视行为，最终促进自由开放社会的发展。pi将设计和传播易于使用的浏览器扩展和插件，这些扩展和插件将能够帮助最终用户检测违反内容中立性的行为。教育是这个奖项的重要组成部分。从这个项目中获得的见解和工具将被整合到当前的本科和研究生课程中。

Millions of users are accessing a portion of billions of Web pages and other content on the Internet on a daily basis. While the networking research community and the public in general are well-aware of the net neutrality problem, i.e., how to develop regulatory policies and auditing mechanisms to prevent Internet Service Providers from discriminating against various applications, very little effort is invested in enabling content neutrality. In particular, it is not a secret that almost every browsing click we make is collected by either Web- or ISP-based 'information collectors and aggregators', and that our profiles are used for online advertising. Still, no public auditing mechanisms, capable of detecting and informing end users about such practices, exist in this emerging area. Take search engines as another example. How does one know that information available on these services is not biased (or will become biased in the future) for commercial, political, or any other reason? More generally, no public auditing systems are capable of monitoring the scope and effectiveness of advertising or spam campaigns on the Internet. The PIs argue that all these questions fundamentally affect fairness at all levels, information credibility, and user privacy, all of which have significant relevance for the future development of the Internet and beyond.The PIs will build a set of methodologies and tools unified in a system capable of (i) enabling auditing mechanisms for the Web advertising domain, (ii) monitoring search engines? services and revealing their neutrality, and (iii) independently determining a Web site's popularity and checking for the truthfulness of advertised popularity. Further, the PIs plan to deploy the developed system on the PlanetLab wide area network testbed to perform long term monitoring of content neutrality in the Internet.Broader Impact: By informing the public of how online advertisers and search engines behave, the auditing system will enable fair competition and preclude monopolies, oligopolies, or collusions, eventually helping to improve online information credibility and user privacy. Even if no bias nor privacy-violating practices were to exist in today's Internet (unfortunately, the PI's preliminary results show that this is not the case), the auditing system and methodologies will always be a barrier for anyone who would believe that applying such approaches might be done without public knowledge and repercussions. By exposing biased content behavior, the auditing system will be capable of revealing discriminating acts that can happen for commercial, social, political, religious, or any other reason, ultimately facilitating the development of free and open society. The PIs will design and disseminate easy-to-use browser extensions and plug-ins that will be capable of assisting end users in detecting content neutrality violations. Education is an integral part of this award. The insights and tools derived from this project will be integrated into the current undergraduate and graduate curricula.