Balancing Privacy and Information Disclosure in Interactive Record Linkage with Visual Masking
Balancing Privacy and Information Disclosure in Interactive Record Linkage with Visual Masking
复制标题
通过视觉屏蔽平衡交互式记录链接中的隐私和信息披露
DOI：
10.1145/3173574.3173900
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
Han Wang
中科院分区：
文献类型：
--
作者：
E. Ragan;Hye‐Chung Kum;Gurudev Ilangovan;Han Wang
有效使用涉及个人或敏感信息的数据通常需要不同的人访问个人信息，这大大降低了数据存储者的个人隐私，并增加了身份盗窃、数据泄露或社会工程攻击的风险。我们的研究研究了人类决策中个人信息的隐私和效用之间的权衡。使用记录链接场景，本文对不同程度的信息可用性如何影响有效使用个人信息的能力进行了对照研究。我们使用可视化界面来比较人类决策的质量，该界面控制可用的个人信息的数量，并使用可视化标记来突出数据差异。有了这个界面，研究参与者只查看了30%的数据内容，他们的决策质量与那些拥有100%访问权限的人相似。结果表明，在不负面影响效用或人类效率的情况下，大大限制人类决策者可用的个人信息的数量是可能的。然而，研究结果也表明，在涉及个人层面数据的决策中，在对判断质量产生负面影响之前，可以隐藏多少数据是有限的。尽管极端数据隐藏会降低准确性，但研究表明，通过适当的界面设计，即使是完全屏蔽的合法去识别数据也可以做出许多正确的决策（完全屏蔽数据的准确率为74.5%，而完全访问的准确率为84.1%）。因此，当法律要求只允许去标识化的数据访问时，使用设计良好的接口可以显著提高数据的实用性。
Effective use of data involving personal or sensitive information often requires different people to have access to personal information, which significantly reduces the personal privacy of those whose data is stored and increases risk of identity theft, data leaks, or social engineering attacks. Our research studies the tradeoffs between privacy and utility of personal information for human decision making. Using a record-linkage scenario, this paper presents a controlled study of how varying degrees of information availability influences the ability to effectively use personal information. We compared the quality of human decision-making using a visual interface that controls the amount of personal information available using visual markup to highlight data discrepancies. With this interface, study participants who viewed only 30% of data content had decision quality similar to those who had full 100% access. The results demonstrate that it is possible to greatly limit the amount of personal information available to human decision makers without negatively affecting utility or human effectiveness. However, the findings also show there is a limit to how much data can be hidden before negatively influencing the quality of judgment in decisions involving person-level data. Despite the reduced accuracy with extreme data hiding, the study demonstrates that with proper interface designs, many correct decisions can be made with even legally de-identified data that is fully masked (74.5% accuracy with fully-masked data compared to 84.1% with full access). Thus, when legal requirements only allow for de-identified data access, use of well-designed interface can significantly improve data utility.