Copyright © 2018 Authors. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted
use, distribution, and reproduction in any medium, provided the original work is properly cited.
International Journal of Engineering & Technology, 7 (3.33) (2018) 187-190
International Journal of Engineering & Technology
Website: www.sciencepubco.com/index.php/IJET
Research paper
A novel property graph model for knowledge representation on
the Web
Heekyung Moon
1
, Zhanfang Zhao
2
, Jintak Choi
3
, Sungkook Han
1
*
1
Department of Computer Engimeering, College of Engineering,Wonkwang University,
460 Iksandae-ro, Iksan, Korea
2
Department of Computer Engimeering, College of Engineering, Hebei GEO University,
Huai An Road No. 136, Shijiazhuang, Hebei Provincei, China
3
Department of Computer Science & Engineering, College of Information Technology,
Incheon National University, 119 Academy-ro, Yeonsu-gu, Incheon, Korea
*Corresponding author E-mail:skhan@wku.ac.kr
Abstract
Graphs provide an effective way to represent information and knowledge of real world domains. Resource Description Framework (RDF)
model and Labeled Property Graphs (LPG) model are dominant graph data models widely used in Linked Open Data (LOD) and NoSQL
databases. Although these graph models have plentiful data modeling capabilities, they reveal some drawbacks to model the complicated
structures. This paper proposes a new property graph model called a universal property graph (UPG) that can embrace the capability of
both RDF and LPG. This paper explores the core features of UPG and their functions.
Keywords: UPG, LPG, RDF, LOD, property graph, key-value pairs, knowledge graph.
1. Introduction
Resource Description Framework (RDF) and Labeled Property
Graphs (LPG) both provide ways to represent knowledge graph-
ically. But they adopt different approaches to model knowledge
representation. RDF is a standard data model for opening, sharing
and interchanging data on the Web. RDF has been widely used as
the core data model for Linked Open Data (LOD) that enables the
Web of Data. Several comprehensive LOD products containing
plentiful RDF datasets have emerged such as DBpedia and YAGO
[1,2]. Besides, a large number of industry RDF datasets have been
published by numerous researchers, institutions, and companies,
which powerfully contribute to the knowledge sharing on the
Web.
Recently, NoSQL databases are attracting increasing attention,
since they address the limitation of Relational Databases (RDB)
and provide more flexible and available data management for
unstructured data [3-5]. Especially, graph databases based on LPG
have received significant attention on account of the good perfor-
mance in dealing with the complex relationships among the data
[6]. LPG model owns distinctive features, using any number of
key-value pairs to describe the properties of vertices and edges,
which makes property graph more expressive and easy to under-
stand for the human being. LPG model shows stronger expres-
siveness than RDF [7]. In addition, databases based on LPG pro-
vide good performance in query and storage for graph data.
The key-value pair properties of LPG model not only have a pow-
erful capability to describe the intrinsic properties of the data ob-
jects but also can model complex relationships efficiently. How-
ever, the conventional LPG model shows lack of semantic expres-
siveness to realize semantic interoperability of the data in the open
and shared environment of the Web. The data types used in LPG
model are also restricted so that it cannot represent the diverse
data structures. The core property structures of LPG need to be
extended to capture both semantic and structural complexities of
the data. This paper proposes a new property graph model called a
universal property graph (UPG) that can embrace the capability of
both RDF and LPG. This paper also explores the features of UPG
model.
The rest of this paper is structured as follows. Section 2 reviews
the related work. Section 3 presents the definition of UPG model
and analyze its characteristics for data modeling. Section 4
demonstrates the applications of UPG model with the diverse use
cases. Section 5 summarizes the contributions and puts forth the
prospects for further work.
2. Related Work
Graphs provide an effective way to represent information and
knowledge of real world domains. There are two dominant graph
data models, RDF and LPG, popular for the applications of big
datasets such as social networks and LOD. Although RDF is
standard for Web-based knowledge modeling and provides the
foundation of LOD, RDF modeling has been widely criticized for
its awkward structures and semantic interpretations. Especially,
the blank nodes and the reification provoke the serious difficulties
in querying and searching [7-9]. Although RDF reification has
been withdrawn from the normative sections in the latest RDF
Recommendation [10], the expressive capabilities of RDF remain
an unresolved problem [11].