How can we maximize the value of accumulated RDF data? Whereas the RDF data can be queried using the SPARQL language, even the SPARQL-based operation has a limitation in implementing traversal or analytical algorithms. Recently, a variety of database implementations dedicated to analyses on the property graph (PG) model have emerged. Importing RDF datasets into these graph analysis engines provides access to the accumulated datasets through various application interfaces. However, the RDF model and the PG model are not interoperable. Here, we developed a framework based on the Graph to Graph Mapping Language (G2GML) for mapping RDF graphs to PGs to make the most of accumulated RDF data. Using this framework, accumulated graph data described in the RDF model can be converted to the PG model, which can then be loaded to graph database engines for further analysis. This study bridges RDF and PGs and contributes to interoperable management of knowledge graphs, thereby expanding the use cases of accumulated RDF data. Demonstration of the G2G mapping framework is available at https://purl.org/g2gml.
Increasing amounts of scienti c and social data are being published in the form
of the Resource Description Framework (RDF), which presently constitutes a
large open data cloud. DBpedia [
Copyright c 2020 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
Recently, the property graph (PG) model [
In the light of this situation, developing a method to transform RDF into PG
would be highly valuable. One of the practical issues faced by this challenge is
the lack of a standardized PG model. Another issue is that the transformation
between RDF and PG is not straightforward due to the di erences in their
models. In RDF graphs, all information is expressed by triples (node-edge-node),
whereas in PGs, arbitrary information can be contained in each node and edge
as the key-value form. Although this issue was previously addressed on the basis
of prede ned transformations [
In this study[
We provide an overview of the graph to graph mapping (G2G mapping) framework (Figure 1).
In this framework, users describe mappings from RDF to PG in G2GML. According to this G2GML description, the input RDF dataset is converted into a PG dataset. The new dataset can also be optionally saved in speci c formats for loading into major graph database implementations.
G2GML is a declarative language comprising pairs of RDF patterns and PG patterns. The core concept of a G2GML description can be represented by a map from RDF subgraphs, which match speci ed SPARQL patterns, to PG components. 3
4 Main paper to be presented at ISWC 2020
{ Resource to node: In lines 2{4, the RDF resources with type :Person are
mapped into the PG nodes using their IRIs as node IDs.
{ Datatype property to node property: In lines 2{4, the RDF datatype
property :name is mapped onto the PG node property key name. The literal
objects 'Alice' and 'Bob' are mapped onto the node property values.
{ Object property to edge: In lines 5{6, the RDF object property :supervised by
is mapped onto the PG edge supervised by.
{ Resource to edge: In lines 7{12, the RDF resource with type :Email is
mapped onto the PG edge emailed.
{ Datatype property to edge property: In lines 7{12, the RDF datatype
property :year and :attachment are mapped onto the PG edge property year
and attachment. The literal objects 2017 and '01.pdf' are mapped onto
the edge property values.
1 "http://example.org/person1" :person name:Alice
2 "http://example.org/person2" :person name:Bob
3 "http://example.org/person1" -> "http://example.org/person2" :supervised_by
4 "http://example.org/person1" -> "http://example.org/person2" :emailed year:2017 attachment:"01.pdf"
A preceding study on converting existing data into graph data included an
effort to convert relational databases into graph databases [
To the best of our knowledge, this study presents the rst attempt to develop a framework for controlled mapping between RDF and PG. Notably, the designed G2GML is a declarative mapping language. As a merit of the declarative description, we can concentrate on the core logic of mappings. In the sense that the mapping process generates new graph data on the basis of existing graph data, it has a close relation to the semantic inference.
Other mapping frameworks, such as Neosemantics (a Neo4j plugins), propose
a method to convert RDF datasets without mapping de nitions. We observe a
similar discussion in the conversion from the relational model to RDF, where
are two W3C standards, i.e., Direct Mapping [
The prototype implementation of G2G mapping is available on GitHub (https: //github.com/g2glab/g2g) under MIT license, which is written in JavaScript and can be executed using Node.js in the command line. It has an endpoint mode and a local le mode. The local le mode uses Apache Jena ARQ to execute SPARQL queries internally, while the endpoint mode accesses SPARQL endpoints via the Internet. An example of the usage in the endpoint mode is as follows:
$ g2g musician.g2g http://dbpedia.org/sparql where the rst argument is a G2GML description le, and the second argument is the target SPARQL endpoint, which provides the source RDF dataset.
Furthermore, a demonstration site (https://purl.org/g2gml) is available, and the documentation (https://g2gml.readthedocs.io) also includes quick tutorials on how to try G2GML using the Docker image (https://hub.docker. com/r/g2glab/g2g).