Project R-14484

Algorithms for query processing under updates (Research)

In the field of database research, the challenge of query processing in dynamic datasets—those subject to continuous updates—presents a complex problem. The conventional approach involves materializing the query result (Q(D)) and the current database state (D). Upon receiving an update (u), a delta query (∆Q) is computed against both the original database and the update, revealing the necessary changes to transform Q(D) into Q(D + u). Recent advancements propose optimizing this process by altering how query answers are materialized. Instead of maintaining Q(D) as a relational table, alternative data structures are explored, potentially offering more compact representations. This PhD will study the concept of dissociated query processing under updates, comprising an update stage focused on maintaining a query result representation and a retrieval stage for effective result retrieval. The primary objective is to investigate this dissociated update framework for novel query languages, under diverse update policies. In particular in the context of non-local query languages with a restricted form of iteration, update policies that allow to group update, and vector stores.

16 October 2023 - 15 October 2027