=Paper=
{{Paper
|id=Vol-2721/paper591
|storemode=property
|title=G2GML: Graph to Graph Mapping Language for Bridging RDF and Property Graphs
|pdfUrl=https://ceur-ws.org/Vol-2721/paper591.pdf
|volume=Vol-2721
|authors=Hirokazu Chiba,Ryota Yamanaka,Shota Matsumoto
|dblpUrl=https://dblp.org/rec/conf/semweb/ChibaYM20a
}}
==G2GML: Graph to Graph Mapping Language for Bridging RDF and Property Graphs==
https://ceur-ws.org/Vol-2721/paper591.pdf
G2GML: Graph to Graph Mapping Language
for Bridging RDF and Property Graphs
Hirokazu Chiba1 , Ryota Yamanaka2 , and Shota Matsumoto3
1
Database Center for Life Science, Chiba 277-0871, Japan
chiba@dbcls.rois.ac.jp
2
Oracle Corporation, Bangkok 10500, Thailand
ryota.yamanaka@oracle.com
3
Lifematics Inc., Tokyo 101-0041, Japan
shota.matsumoto@lifematics.co.jp
Abstract. 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 ac-
cess 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 Lan-
guage (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 analy-
sis. 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.
Keywords: RDF · Property Graph · Graph Database
1 Introduction
Increasing amounts of scientific and social data are being published in the form
of the Resource Description Framework (RDF), which presently constitutes a
large open data cloud. DBpedia [1] and Wikidata [2] are well-known examples
of such RDF datasets. SPARQL is a protocol and query language that serves
as a standardized interface for RDF data. This standardized data model and
interface enables the construction of integrated graph data. However, the lack of
an interface for graph-based analysis and performant traversal limits use cases
of the graph data.
Copyright c 2020 for this paper by its authors. Use permitted under Creative Com-
mons License Attribution 4.0 International (CC BY 4.0).
Recently, the property graph (PG) model [3,4] has been increasingly attract-
ing attention in the context of graph analysis. Various graph database engines,
including Neo4j [5], Oracle Database [6], and Amazon Neptune [7] adopt this
model. These graph database engines support algorithms for traversing or an-
alyzing graphs. However, few datasets are published in the PG model and the
lack of an ecosystem for exchanging data in the PG model limits the application
of these powerful engines.
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 differences 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 predefined transformations [8], users still cannot control the mapping for their
specific use cases.
In this study[9]4 , we redefine the PG model incorporating the differences
in existing models and propose serialization formats based on the data model.
We further propose a graph to graph mapping framework based on the Graph
to Graph Mapping Language (G2GML). Employing this mapping framework,
accumulated graph data described in RDF can be converted into PGs, which
can then be loaded into several graph database engines for further analysis.
2 Overview
We provide an overview of the graph to graph mapping (G2G mapping) frame-
work (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 specific 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 specified SPARQL patterns, to PG
components.
3 Examples
Figure 3 shows the minimal example of G2G mapping from RDF data (Fig-
ure 2) to PG data (Figure 4), representing the following five types of typical
mapping. Here, this PG data is represented in a general PG format we defined
previously [10].
4
Main paper to be presented at ISWC 2020
Fig. 1. Overview of mapping from RDF to PG
1 @prefix : .
2 :person1 a :Person ;
3 :name ’Alice’ .
4 :person2 a :Person ;
5 :name ’Bob’ .
6 :person1 :supervised_by :person2 .
7 [] a :Email ;
8 :sender :person1 ;
9 :receiver :person2 ;
10 :year 2017 ;
11 :attachment ’01.pdf’ .
Fig. 2. Example of input RDF data
1 PREFIX :
2 (p:person {name:n})
3 ?p a :Person .
4 ?p :name ?n .
5 (p1:person)-[:supervised_by]->(p2:person)
6 ?p1 :supervised_by ?p2 .
7 (p1:person)-[:emailed {year:y, attachment:a}]->(p2:person)
8 ?f a :Email ;
9 :sender ?p1 ;
10 :receiver ?p2 ;
11 :year ?y .
12 OPTIONAL { ?f :attachment ?a }
Fig. 3. Example of G2G mapping definition
– 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 prop-
erty :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 prop-
erty :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"
Fig. 4. Example of output PG data
Fig. 5. Example of output PG data (Visualization)
4 Related Work
A preceding study on converting existing data into graph data included an ef-
fort to convert relational databases into graph databases [11]. However, given
that RDF has prevailed as a standardized data model in scientific communities,
considering mapping based on the RDF model is crucial. The interoperability of
RDF and PG [8,12,13,14] has been discussed, and efforts were made to develop
methods to convert RDF into PG [15,16]. However, considering the flexibility
regarding the type of information that can be expressed by edges in property
graphs, a novel method for controlling the mapping is necessary.
To the best of our knowledge, this study presents the first attempt to de-
velop a framework for controlled mapping between RDF and PG. Notably, the
designed G2GML is a declarative mapping language. As a merit of the declara-
tive 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 definitions. We observe a
similar discussion in the conversion from the relational model to RDF, where
are two W3C standards, i.e., Direct Mapping [17] and R2RML [18].
5 Availability
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 file mode. The local file 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 first argument is a G2GML description file, 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).
References
1. Lehmann, J., Isele, R., Jakob, M., Jentzsch, A., Kontokostas, D., Mendes, P. N.,
... Bizer, C.: DBpedia–a large-scale, multilingual knowledge base extracted from
Wikipedia. Semantic Web, 6(2), 167-195. (2015)
2. Vrandečić, D., Krötzsch, M.: Wikidata: a free collaborative knowledgebase. Com-
munications of the ACM, 57(10), 78-85. (2014)
3. Angles, R., Gutierrez, C.: An introduction to Graph Data Management. arXiv
preprint arXiv:1801.00036 (2017)
4. Angles, R., Arenas, M., Barceló, P., Hogan, A., Reutter, J., Vrgoc, D.: Founda-
tions of Modern Query Languages for Graph Databases. ACM Computing Surveys
(CSUR), 50(5), 68. (2017)
5. The Neo4j Graph Platform. https://neo4j.com/
6. Oracle Database Property Graph. https://www.oracle.com/goto/propertygraph
7. Amazon Neptune. https://aws.amazon.com/neptune/
8. Hartig, O.: Reconciliation of RDF* and property graphs. arXiv preprint
arXiv:1409.3288 (2014)
9. Chiba, H., Yamanaka, R., Matsumoto, S.: G2GML: Graph to Graph Mapping Lan-
guage for Bridging RDF and Property Graphs. Proceedings of the 19th International
Semantic Web Conference (2020)
10. Chiba, H., Yamanaka, R., Matsumoto, S. Property Graph Exchange Format. arXiv
preprint arXiv:1907.03936 (2019)
11. De Virgilio, R., Maccioni, A., Torlone, R.: Converting relational to graph databases.
In First International Workshop on Graph Data Management Experiences and Sys-
tems, p. 1. ACM (2013)
12. Angles, R., Thakkar, H., Tomaszuk, D.: RDF and Property Graphs Interoperabil-
ity: Status and Issues. Proceedings of the 13th Alberto Mendelzon International
Workshop on Foundations of Data Management (2019)
13. Das, S., Srinivasan, J., Perry, M., Chong, E. I., Banerjee, J.: A Tale of Two Graphs:
Property Graphs as RDF in Oracle. In EDBT, pp. 762–773 (2014)
14. Thakkar, H., Punjani, D., Keswani, Y., Lehmann, J., Auer, S.: A Stitch in Time
Saves Nine–SPARQL querying of Property Graphs using Gremlin Traversals. arXiv
preprint arXiv:1801.02911 (2018)
15. Tomaszuk, D.: RDF data in property graph model. In Research Conference on
Metadata and Semantics Research, pp. 104–115 (2016)
16. De Virgilio, R.: Smart RDF data storage in graph databases. In 2017 17th
IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing, pp.
872–881. IEEE (2017)
17. A Direct Mapping of Relational Data to RDF, W3C Recommendation 27 Septem-
ber 2012 https://www.w3.org/TR/r2rml/
18. R2RML: RDB to RDF Mapping Language, W3C Recommendation 27 September
2012 https://www.w3.org/TR/r2rml/