Query Hints in a Knowledge Graph Query Engine for Decentralized Data
Keywords: Linked Data, Querying, RDF, SPARQL, Webdevelopment, decentralization, Javascript, Typescript
Promotors: Ruben Verborgh, Ruben Taelman
Students: max 1
Problem
SPARQL query engines are tools to query over RDF-based Knowledge Graphs.
Such will convert SPARQL query strings into an executable query plan after a query planning phase.
Many query engines allow query hints to be given to the query engine to force the query plan to take a specific form, in case the query writer has more information than the query planner.
So far, such query hints have only been designed to work for Centralized Knowledge Graphs, and it is not known yet how these will behave for the more recently upcoming Decentralized Knowledge Graphs.
These are Knowledge Graphs that are spread over the Web in different data sources, and are linked to each other.
Goal
In this thesis, you will implement query hints into the Typescript-based Comunica SPARQL query engine, that enables querying over Decentralized Knowledge Graphs.
Next to implementing query hints for traditional centralized query engines (physical join plan, join order, ...), you will develop novel query hints that can guide the processing of decentralized data, such as informing the query engine which links to different sources need to be followed, and in what order links need to be followed.