Supporting Query by Content for Time Series Data in Relational Database Management Systems

Seminar Slides: Download (PDF)

When: Monday, November 25, 2013
Where: PGH 563
Time: 11:00 AM

Speaker: Prof. Ines F. Vega-Lopez, University of Sinaloa, Mexico

Host: Prof. Carlos Ordonez

Similarity search, or query by content, is an important operation for time series databases. While the research community has been quite active in this area, we are yet to see full support for these operations from database systems. In consequence, ad-hoc queries and file repositories are used to develop applications that require this kind of functionality. It is well known that using the file system as the backbone for a database has its drawbacks. Therefore, our work focuses on providing formal models and tools for properly and efficiently storing, querying, and analyzing this type of data. The goal is to incorporate these models and tools into database systems. In this presentation, I will describe the challenges we must face when managing time series data in a database system. I will illustrate these challenges by using ECG data, an example from the medical domain. ECG data is abundant (large quantities of ECG signals are generated everyday by medical centers) and complex (e.g., variable length and sampling ratio). In addition, the social impact of providing database technology for supporting large-scale ECG data analysis is evident.  Finally, I will present our work on extending tried and tested relational database management systems to support complex operations for sequence classification on ECG data. The efficiency of this approach will be compared to the use ad-hoc queries and file repositories.

Bio:
Ines F. Vega-Lopez is a professor and head of the graduate program at the School of Computer Science at the University of Sinaloa, Mexico. He received a PhD in computer science from the University of Arizona in 2004, under Dr. Bongki Moon’s supervision. Ines’ current research interest includes developing techniques for efficiently storing and querying of sensor data. He is also interested in mining medical and conventional data. In the past, Ines has worked with temporal and spatio-temporal databases.