Improving Users' Access to Quality Data

点击翻译按钮获取中文摘要

In today's era of big data, users have access to more data than ever. Despite the availability of data, many users still have difficulty finding reliable, high-quality data for their applications. Some crucial factors that impact how users access and use data include: the proliferation of poor quality data, data privacy regulations, and the availability of large-scale and heterogeneous datasets. Over the course of the next five years, I will investigate some of the new challenges that database users face when trying to obtain high-quality data. I will focus on three challenges that I believe are particularly critical: querying and exploring databases, cleaning and repairing data, and protecting sensitive and private information. My research primarily concerns database users, i.e., users who directly interact with database systems, including software developers, data analysts, and data scientists. However, my research outcomes will ultimately benefit the general public through the products and services that the database users provide. The goal of my research program is two-fold: I intend to formulate problems related to the above challenges and I will develop practical solutions and algorithms that address these problems. Both aspects of my research goal (identifying problems and solutions) are equally important, although their outcomes and benefits are very different. Problem formulation creates a framework for developing and evaluating solutions; however, problems are important to database research in themselves because they generate new questions and research directions for other researchers to pursue. I expect that the resulting solutions will be used to either build new systems or to improve existing ones. Ultimately, this research will help software developers build reliable software systems. It will also help data scientists perform accurate and efficient data analysis, which will, in turn, benefit general users, who will have access to better quality data. In this proposal, I will describe my main objectives. I intend to work with my future students to achieve the following research objectives: (1) integrating data cleaning with querying databases, (2) designing techniques to assess the quality of private data, and (3) improving Query by Example (QBE) using query logs. The nature of this research is both theoretical and experimental. In terms of the theoretical aspects, I will use formal models for problem definitions, and I will apply different techniques (machine learning, statistical analysis, and probabilistic models) to solve these problems. On an experimental or pragmatic level, I will demonstrate the usefulness of my proposed solutions in real-world scenarios by gathering empirical data through user studies and surveys.