Modular materialisation of Datalog programs

摘要

Answering queries over large datasets extended with Datalog rules plays a key role in numerous data management applications, and it has been implemented in several highly optimised Datalog systems in both academic and commercial contexts. Many systems implement reasoning via materialisation, which involves precomputing all consequences of the rules and the dataset in a preprocessing step. Some systems also use incremental reasoning algorithms, which can update the materialisation efficiently when the input dataset changes. Such techniques allow queries to be processed without any reference to the rules, so they are often used in applications where the performance of query answering is critical.