Scalable Dynamic Query Evaluation for Data Science (Research)
This research is embedded in the general research theme of "Data Science Processing inside the Data Engine" where the aim is to study how computational data science tasks that are typically executed outside of a database management system, such as machine learning and real-time analytics, can be executed inside of a data engine. Specifically, my research focuses on the problem of Dynamic Query Evaluation which is the problem where a given database query Q has to be evaluated against a database that is constantly updated. In this setting, when database db is updated to database (db + u) under update u, the objective is to efficiently compute Q(db + u), taking into consideration that Q(db) was already evaluated and re-computations could be avoided. I will study how dynamic query evaluation behaves in the presence of features such as linear algebra operations, general aggregations, and recursion --- which all occur in the machine learning context.
Period of project
01 February 2022 - 31 January 2024