Forming better database queries at heart of NSF research project

A UT Arlington computer scientist is helping design a system that will ask better questions when querying databases and lead to improved decision-making in our data-driven society.

Chengkai Li, an associate professor in the Computer Science & Engineering Department, has been awarded a $241,778 National Science Foundation grant for his proposal about "perturbation analysis of database queries." He is teaming with Duke University and Stanford University on the joint grant, which is worth more than $1.2 million.

Perturbation analysis studies how tweaks of database query templates and parameters affect query results.

"In this project, the focus of database research is switched from answering questions to questioning answers and even the questions themselves," said Li, whose primary area of expertise is the field of big data management and mining. "What the research aims to do is reveal how queries affect the robustness and objectivity of decisions. It helps decision makers identify 'good' queries that will influence their decisions."

Khosrow Behbehani, dean of the College of Engineering, said Li's research has so many applications in today's world.

"In the age of big data, Dr. Li's research has significant applications. It could be used in making public policy, urban planning, product development, and smart health care," Behbehani said. "In today's world successful economies are based on robust data and intelligent analysis of the data. The future of new discoveries from any databases, and in particular giant databases, rests on the soundness of the inquiries from the database "

Eventually, Li said the team will create a menu of courses, seminars and workshops targeting data-driven decision makers and educating them on how to construct more sensible queries.

"A poor choice of queries could yield decisions that are wrong or misguided," Li said. "And asking those poor questions may be intentional or unintentional. Database research in the past has focused on how to answer queries. There hasn't been much done on how the queries impact decision-making or how to formulate good queries. This research looks at that."

Provided by University of Texas at Arlington