Research Projects:

Exploring Ontological Rules for Reasoning and Query Answering on Databases

This project centers on the exploration of rule-based languages for reasoning and query answering within the context of relational databases. The primary focus is on the chase, a fundamental and versatile tool that plays a crucial role not only in reasoning and query answering but also as a solution to various other problems within databases. The chase procedure is a vital aspect of this research, as it takes a database and a set of constraints and iteratively refines the database according to those constraints. The importance of the chase extends beyond mere constraint satisfaction; it serves as a bridge to understanding complex relational structures and enhancing the efficiency of query processing. The project also delves into the applications and extensions of rule-based languages, such as weakly-sticky Datalog±, providing insights into their behavior and optimization. By studying the generalized stickiness of the chase and proposing algorithms for query answering, the research contributes to a broader understanding of rule-based reasoning within relational databases. Together, these elements form a cohesive exploration of the chase and its multifaceted applications. By emphasizing the chase’s role in rule-based languages, the project offers valuable insights into the dynamics of reasoning and query answering in relational databases. It represents a significant step towards advancing the understanding of this essential tool and its broader applications within the field of databases.

Updated August 21, 2023

Building Trust in Data: A Holistic Approach to Privacy, Fairness, and Data Quality

This project embodies a forward-thinking approach to data quality management, weaving together the essential principles of quality, privacy, fairness, and responsible stewardship. Recognizing the complex interplay between these facets, the research crafts a unified framework that prioritizes ethical data handling. At the heart of the project is a commitment to data quality, exploring innovative techniques to ensure the integrity and cleanliness of data. This includes the development of privacy-aware data cleaning models and the integration of diversity constraints in data anonymization, aligning with modern privacy standards. Beyond quality and privacy, the project takes a decisive stance on fairness, striving to eliminate bias and promote equitable data processing. By embedding fairness considerations into the data management process, the research contributes to a more transparent and unbiased data ecosystem. Together, these components form a cohesive vision for data quality management. By harmonizing quality assurance with privacy protection and fairness, the project sets a new standard for ethical data practices. It offers a practical and comprehensive solution that resonates with the evolving demands of data stewardship, reflecting a profound commitment to building trust, responsibility, and quality in data handling.

Updated April 11, 2023

Unified Data Exploration: Provenance Insights and Query Recommendations

This project represents a unified approach to enhancing data access and utility, focusing on two interconnected aspects of database management: Provenance Exploration and Intelligent Query Recommendation. Data provenance, referring to the origin and process leading to the creation of data, has been a complex and vital area of research. The project introduces innovative techniques for the efficient generation and summarization of provenance information in relational databases. Novel summarization schemes are developed to provide a more user-friendly representation of data’s origin, making provenance information more accessible and relevant. Building on this foundation, the project also explores the development and implementation of data-driven techniques for SQL query recommendation. By harnessing the power of large-scale query workloads, the research introduces methods to enhance query recommendations and predictions. A standout feature is a workload-aware, deep-learning approach, along with collaborative filtering techniques like the Skyrec Summary method. These advancements enhance user interaction and support in SQL and scientific databases. Together, these two aspects form a cohesive approach to data access and utility. By focusing on provenance exploration and intelligent query recommendation, the project addresses key challenges in database management, offering practical solutions that resonate with the evolving landscape of data access. The integration of these themes reflects a commitment to advancing database technology and user experience, setting a new standard for data access and exploration.

Updated November 10, 2022