Foundations of RDF⋆ and SPARQL⋆ - Leitura de Artigo

Hartig, Olaf. “Foundations of RDF⋆ and SPARQL⋆ (An Alternative Approach to Statement-Level Metadata in RDF).” AMW (2017).

 

The term statement-level metadata refers to a form of data that captures information about another piece of data representing a single statement or fact. A typical example are so called edge properties in graph databases; such an edge property takes the form of a key-value pair that captures additional information about the relationship represented by the edge with which the key-value pair is associated

 

While the Resource Description Framework (RDF) [1] presents another graph-based approach to represent statements about entities and their relationships, its triple-based data model does not natively support the representation of statement-level meta-data. To mitigate this limitation the RDF specification introduces the notion of RDF reification which can be used to provide a set of RDF triples that describe some other RDF triple  

 

The example highlights two major shortcomings of RDF reification: First, adding four reification triples for every reified triple is inefficient for exchanging RDF data. Second, writing queries to access statement-level metadata is cumbersome because any metadata-related (sub)expression in a query has to be accompanied by another subexpression to match the corresponding four reification triples.

 

That is, in the extended language, called SPARQL*, triple patterns may also be nested, which gives users a query syntax in which accessing specific metadata about a triple is just a matter of mentioning the triple in the subject (or object) position of a metadata-related triple pattern.

 

As a basis for the following definitions, we assume pairwise disjoint sets I (all IRIs), B(blank nodes), and L(literals). As usual, a tuple in (I∪B)×I×(I∪B∪L) is an RDF triple and a set of RDF triples is an RDF graph [1]. RDF* extends the notion of such triples by allowing for triples that have another triple in its subject or its object position. Such nesting may be arbitrarily deep. The following definition captures this notion.

 
Definition 1. An RDF* triple is a 3-tuple that is defined recursively as follows:
1. Any RDF triple t ∈ (I∪B) ×I×(I∪B∪L) is an RDF* triple; and
2. Given RDF* triples t and t′, and RDF terms s ∈ (I ∪ B), p ∈ I and o ∈ (I ∪ B ∪ L), then the tuples (t,p,o), (s,p,t) and (t,p,t′) are RDF* triples.

 

We recall that the basic building block of SPARQL queries is a basic graph pattern (BGP); that is, a finite set of triple patterns, where every triple pattern is a tuple of the form (s,p,o) ∈ (V ∪ I ∪ L) ×(V ∪ I) ×(V ∪ I ∪ L) with V being a set of query variables that is disjoint from I, B, and L, respectively. SPARQL? extends these patterns by adding the possibility to nest triple patterns within one another.

 
Definition 2. A triple* pattern is a 3-tuple that is defined recursively as follows:
1. Any triple pattern t ∈ (V ∪ I ∪ L) × (V ∪ I) × (V ∪ I ∪ L) is a triple* pattern; and
2. Given two triple* patterns tp and tp', and s ∈ (V ∪ I ∪ L), p ∈ (V ∪ I) and o ∈(V ∪ I ∪ L), then (tp,p,o), (s,p,tp) and (tp,p,tp') are triple* patterns.

 

Before defining a query semantics of SPARQL*, we recall that the semantics of SPARQL is defined based on the notion of solution mappings [2,7], that is, partial mappings μ: V → (I ∪ B ∪ L). For SPARQL* we extend this notion to so-called solution* mappings that may bind variables not only to IRIs, blank nodes, or literals, but also to RDF* triples. Hence, a solution* mapping η is a partial mapping η : V →(T* ∪ I ∪ B ∪ L).

 

A triple-to-ID mapping id is an injective function id : T*→ (I ∪ B).