Named Graphs. Enabling Provenance and Digital Signatures in RDF

Named Graphs: Enabling provenance and digital signatures in RDF

Jonas Weber

Universität Freiburg

ABSTRACT

This paper provides an overview over the proposal ’Named Graphs’ by Jeremy Caroll, Christian Bizer, Pat Hayes and Patrick Stickler. Named graphs are an extension of RDF adding the possibility to assign names to different subsets of graphs. These names can then be used to treat the triples encoding a graph as a resource, making it possible to store metadata for it. Well-defined semantics and serializations are crucial to provide an open and extensible system to ’talk’ about graphs. Interesting applications of this include quoting and cryptographically secure signing of named graphs in semantic web, providing provenance information and trust.

1. INTRODUCTION

RDF provides an adaption of the mathematical graph model into applied computer science by defining protocols and formats for the serialization of graphs. The RDF model is based on triples, consisting of subject, predicate and object. Only a small set of semantics is defined by the RDF standard itself, including a method to extend these predefined semantics.

The original specification of RDF graphs and corresponding semantics by Hayes and contributors from 2004 ([7]) did not provide any way to attach a name to an RDF graph. Identifiers in the form of an IRI^[1] are used to identify resources, but graphs of resources were not considered resources by themselves.

However, it is crucial to have the ability to ’talk’ about a graph by itself to enable new and innovative approaches for the semantic web. We need a graph as a resource by itself if we want to quote it. Furthermore, we also need to have a fixed set of triples being the elements of a graph to be able to sign it.

In the paper ’Named Graphs’ ([3]) the authors Caroll, Bizer, Hayes and Stickler introduced the notion of a name for a whole graph. Applications include storing governance

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.

LINKED-DATA ’14/15 Uni Freiburg Not published. information about a graph (for example the author or institutions warranting the correctness of the information) and digitally signing such graphs to prove authorship, but also to be able to reference a graph when writing comments about it.

Named graphs are a semantic extension to RDF. By redefining axioms for a larger domain existing derived properties such as OWL semantics can be applied to this extension as well.

Firstly, the abstract syntax for named graphs is introduced. After that, the additional semantics built on top of the existing framework are explained. Then, a short overview over serialization formats and implementations in popular systems follows. The next section shows how the problems discussed above can be solved by named graphs. In the end related work is discussed and a conclusion is done.

RDF graphs may contain blank nodes, i.e. nodes that are not identified by IRIs but by names that are local to the graph serialization. For simplicity, in this paper equivalence of graphs is assumed to ignore blank node names, as it is described in RDF concepts ([14]). Renaming of blank node names is assumed to happen automatically when two graphs are merged.

2. ABSTRACT SYNTAX

RDF graphs are defined using a mathematical model of a graph. The concepts defined in RDF semantics² are just basic framework semantics that make it possible to extend it freely.

The first important distinction the authors made is between ’named graphs and the RDF graph that the named graph encodes or represents’ ([3]). This means that the graph itself does not have a name, but another entity which references it.

Formally speaking, an entity of type named graph (for example ng) has two properties which are describable with functions:

name(ng) denotes the name of the graph. This value is an IRI^[3].

The other function is rdf graph(ng), which denotes the (sub)graph that is named.

According to RDF semantics any number of triples forms a graph, which is a subgraph of a graph containing at least those triples. The range of rdf graph is all subgraphs.

[...]

^[1] Internationalized Resource Identifier, an expansion to Unicode characters of URIs

^[2] [8]

^[3] The authors of [3] specified n as an URI, but this was amended later to IRIs in [8]