TC:Large:Collaborative Research:Anonymizing Textual Data and its Impact on Utility

数据保护法免除了无法单独识别的数据，这导致了匿名化研究的爆炸性增长。不幸的是，目前的去身份和匿名化技术对隐私和保密的风险控制得有多好还没有被很好地理解。匿名数据对现实世界的应用程序也没有用处。该项目从三个方面解决匿名化问题：1)文本数据，即使删除了显式标识(姓名、日期、位置)，也可能包含高度可识别的信息。例如，印第安纳州患者护理网络(INPC)的一份主诉报告样本发现了几个“幻肢痛”的例子。截肢者可以在视觉上识别，但HIPAA安全港规则没有将此列为“识别信息”。任何明确列出所有类型的识别数据的政策都可能失败。通过与计算机科学和语言学的共同努力，该项目正在开发新的方法，在保留意义的同时删除文本中的特定细节，消除这种高度可识别的信息，而不是事先知道什么是可识别的。2)目前的匿名化研究基于未经证实的可识别性测量。通过对合成数据(但基于真实医疗数据)的重新识别挑战，该项目正在评估这些措施的有效性。跨学科的学生团队被要求提出挑战性问题--使用假设的医疗数据匿名的数据--并被要求对个人的健康信息做出(假设的)推断。结果可以用来校准不同匿名化措施的有效性。3)匿名化数据的效用一直是研究中关注的问题：匿名化数据是否提供可信的研究结果？通过与金赛研究所和普渡大学护理学院的医疗研究合作，该项目正在比较对原始数据的分析和对匿名数据的分析，并评估匿名类型对研究结果的影响。一个相关的问题正在决定对数据收集的影响：如果个人知道数据将被匿名，他们的回应是否会更加坦率？结果扩大了可以对匿名数据进行研究的范围，同时确保研究人员知道何时需要访问可单独识别的数据(附带限制和保障)。通过这些任务，该项目正在提高我们利用我们现在收集的丰富数据造福社会的能力，同时确保个人隐私受到保护。有关更多信息，请参见项目网站URL:http://projects.cerias.purdue.edu/TextAnon

Data Protection laws that exempt data that is not individuallyidentifiable have led to an explosion in anonymization research.Unfortunately, how well current de-identification and anonymizationtechniques control risks to privacy and confidentiality is not wellunderstood. Neither is the usefulness of anonymized data for real-worldapplications. The project addresses anonymization on three fronts:1) Textual data, even when explicit identifiers are removed (names,dates, locations), can contain highly identifiable information. Forexample, a sample of chief complaint fields from the Indiana Networkfor Patient Care (INPC) found several instances of "phantom limbpain". Amputees can be visually identifiable, but the HIPAA SafeHarbor rules do not list this as "identifying information". Anypolicy explicitly listing all types of identifying data is likely tofail. Through a joint effort with computer science and linguistics,the project is developing new methods to remove specific details fromtext while preserving meaning, eliminating such highly identifiableinformation without a priori knowledge of what would be identifying.2) Current anonymization research is based on unproven measures ofidentifiability. Through a re-identification challenge on syntheticdata (but based on real healthcare data), the project is evaluatingthe efficacy of these measures. Interdisciplinary teams of studentsare given challenge problems - anonymized data with hypotheticalhealthcare data - and asked to make (hypothetical) inferences abouthealth information of individuals. The results can be used tocalibrate the effectiveness of different anonymization measures.3) The utility of anonymized data has been a concern among research:Does anonymized data provide credible research results? By partneringwith healthcare studies at the Kinsey Institute and Purdue UniversitySchool of Nursing, the project is comparing analyses on original datawith analyses on anonymized data, and evaluating the impact of typesof anonymization on research results. A related issue is determiningthe impact on data collection: Are individuals more candid in theirresponses if they know data will be anonymized? Outcomes are broadeningthe scope of research that can be performed on anonymized data, whileensuring that researchers know when access to individually identifiabledata (with attendant restrictions and safeguards) is needed.Through these tasks, the project is advancing our ability to utilizethe wealth of data we now collect for the benefit of society, whileensuring individual privacy is protected.For further information see the project web site at the URL:http://projects.cerias.purdue.edu/TextAnon