5.4. AMBER: a SPARQL querying engine

The proposed AMBER (Attributed Multigraph Based Engine for RDF querying) contains two different stages: (1) an offline stage, during which RDF data are transformed into multigraph G and then a set of index structures I are constructed that capture the necessary information contained in G; (2) an online stage, during which a SPARQL query is transformed into a multigraph Q, and then by exploiting the subgraph matching techniques along with the already built index structures I, the homomorphic matches of Q in G are obtained.

Given a multigraph representation Q of a SPARQL query, AMBER decomposes the query vertices U into a set of core vertices U_c and satellite vertices U_s. Intuitively, a vertex u ∈ U is a core vertex if the degree of the vertex is more than one; on the contrary, a vertex u with degree one is a satellite vertex. For example, in Figure 5.2(c), U_c = {u₁, u₃, u₅} and U_s = {u₀,u₂, u₄, u₆}. Once decomposed, we run the sub-multigraph...