Data Integration: A perpetually evolving challenge for new research perspectives

Jury

• Amel Bouzeghoub, Professor, Telecom SudParis, Samovar Laboratory
• Zoubida Kedad, Professor, Versailles Saint-Quentin-en-Yvelines University, Paris-Saclay University, DAVID laboratory
• Mohan C. Mohan, Distinguished Professor, IBM/Hong Kong Baptist University.
• Paolo Papotti, Associate Professor, Eurecom.
• Pascal Poncelet, Professor, Montpellier University, LIRRM Laboratory.
• Genoveva Vargas-Solar, Full-time researcher, CNRS, LIRIS Laboratory.

Abstract

Data is everywhere and can be produced and stored from everyday scenarios. The structures used for storing data are heterogeneous, continuously evolving, and consumed by applications according to different patterns. It is demonstrated that most of the value of any application is in the ability to access useful and reliable data. From data processing, we can extract value and insights. In addition, the more different and heterogeneous the data are, the more challenging it is to analyze, understand, and make predictions on top of them. In my work, I summarize part of my research contributions to the field of heterogeneous data integration. First, I present the classical data integration scenarios on NoSQL databases. Then I address the challenge of developing new integration techniques conceived for feeding Artificial Intelligence algorithms. The illustration will focus on applications arising from multiple domains used as case studies during the development of the research ideas: smart cities, carbon storage, user profiling, music recognition, and medical data integration. Finally, I investigate the application of data integration techniques to new challenges and the development of new prospective oriented in data for AI.