From: VEDAS: an efficient GPU alternative for store and query of large RDF data sets
Representation | Detail | Advantage | Disadvantage |
---|---|---|---|
Relational | Store a list of triples with indices e.g. RDF-3X, TriAD | Simple and many techniques from relational database can be applied | Intensive computation for join the related data |
Graph | Store subjects and objects as nodes and predicate as edges e.g. gStore | Natural way to retrieve link data | Scaling limitation on in-memory database. Hard to parallelize for a SIMD architecture |
Matrix | Use a sparse square matrix to represent the RDF e.g. TripleBit, gSMat, MAGiQ | Reuse the existing matrix kernel on various platforms | Fixed size, complicate to represent the multigraph |