III: Small: A Theoretical Framework for Practical Entity Resolution in Network Data

在信息超载和大数据时代，对数据进行分析、保护、排序和利用是迫切需要的。这些数据中的大部分都是固有的关系数据；因此，理解利用关系属性的好处、挑战和潜在危险是至关重要的。集成、清理和链接关系数据需要匹配和解析数据中的引用。同时，匹配和链接也带来了很大的隐私风险。提出的工作发展了对网络数据中实体解析的理论理解，其目标是开发工具和方法，可以告诉我们在不同设置中解析数据的难易程度。利用这一理论，新的实体解析算法将被开发出来，保证精度，并将实体解析扩展到大规模数据源。这些研究结果将为个人、行业和政府提供更明智的数据共享和使用决策。准确分析网络数据对科学、医学和国家安全至关重要。无论是研究社会经济趋势，整合来自大型微阵列的数据，分析有组织犯罪或恐怖主义网络，还是挖掘企业不当行为的金融数据，准确的网络数据及其相关统计数据都至关重要。与此同时，了解实体解析如何影响隐私保障，并教育公众了解发布身份信息的影响，也同样重要。

In an era of information overload and big data, there is a pressing need to analyze, protect, prioritize and utilize data. Much of this data is inherently relational; thus, it is crucial to understand the benefits, challenges and potential hazards of exploiting the relational properties. Integrating, cleaning, and linking relational data requires matching and resolving references in the data. At the same time, matching and linking pose significant privacy risks. The proposed work develops a theoretical understanding of entity resolution in network data with the goal of developing tools and methods which can tell us how easy or difficult it will be to resolve data in different settings. Making use of the theory, new entity resolution algorithms will be developed with accuracy guarantees and for scaling entity resolution to large-scale data sources. These research results will enable more informed data sharing and usage decisions by individuals, industry, and government. Accurate analysis of network data is of utmost importance to science, medicine and national security. Whether studying socioeconomic trends, integrating data from large microarrays, analyzing organized crime or terrorist networks, or mining financial data for corporate misconduct, accurate network data, and its associated statistics, are crucial. At the same time, understanding how entity resolution effects privacy guarantees, and educating the public about the impact of releasing identifying information, is equally important.