Local First RDF and Knowledge Graphs
Keywords: Linked Data, Querying, RDF, SPARQL, Webdevelopment, decentralization
Promotors: Ruben Verborgh, Ruben Taelman
Students: max 1
Problem
The local-first paradigm improves application responsiveness by allowing data to be modified locally before synchronizing changes in the background. This ensures a seamless user experience, even with poor or no internet connectivity. Under the hood, technologies like Conflict-Free Replicated Data Types (CRDTs) and Operational Transforms (OTs) are used to merge diverging data structures. Various CRDTs exist for different data types, such as text, lists, and sets.
RDF datasets, structured as sets of triples (subject - predicate - object) or quads (which include a graph), could benefit from CRDT-based synchronization. A set-based CRDT could be used to merge RDF datasets, but it would not support collaborative editing of literals, such as strings. To enable real-time collaboration on RDF literals, a text-based CRDT would be required. However, there is currently no systematic analysis of which CRDTs are best suited for RDF synchronization, considering both functional and performance-related factors.
Goal
This thesis will conduct a comparative study on the applicability of different CRDT types for RDF datasets. Specifically, the student will:
- Analyze and compare different CRDT types based on their suitability for RDF synchronization.
- Evaluate both functional and non-functional requirements, including memory overhead, client and server load, and semantic relevance to RDF.
- Demonstrate how CRDT-aware clients can collaboratively edit the same RDF dataset in a seamless and efficient manner.
By identifying the most suitable CRDTs for RDF synchronization, this research will contribute to enabling offline-capable, collaborative RDF applications while maintaining consistency across distributed clients.