Exploration and Visualization of Big Graphs

The DBpedia Case Study

Enrico G. Caldarola

1,2

, Antonio Picariello

, Antonio M. Rinaldi

1,3

and Marco Sacco

Department of Electrical Engineering and Information Technologies, University of Naples Federico II, Napoli, Italy

Institute of Industrial Technologies and Automation, National Research Council, Bari, Italy

IKNOS-LAB Intelligent and Knowledge Systems, University of Naples Federico II, LUPT 80134, Napoli, Italy

Institute of Industrial Technologies and Automation, National Research Council, Milano, Italy

Keywords:

Graph Database, Big Data, NoSQL, Data Visualization, DBpedia, Neo4J.

Abstract:

Increasingly, the data and information visualization is becoming strategic for the exploration and explanation

of large data sets. The Big Data paradigm pushes for new ways, new technological solutions to deal with the

big volume and the big variety of data today. Not surprisingly, a plethora of new tools have emerged, each of

them with pros and cons, but all espousing the cause of ”Bigness of Data”. In this paper, we take one of this

emerging tools, namely Neo4J, and stress its capabilities in order to import, query and visualize data coming

from a big case study: DBpedia. We will describe each step in this study focusing on the used strategies

for overcoming the different problems mainly due to the intricate nature of the case study and its volume.

We confront with both the intensional schema of DBpedia and its extensional part in order to obtain the best

result in its visualization. Finally, an attempt to deﬁne some criteria to simplify the large-scale visualization of

DBpedia will be made, providing some examples and considerations which have arisen. The ultimate goal of

this work is to investigate techniques and approaches to get more insights from the visual representation and

analytics of large graph databases.

1 INTRODUCTION

Nowadays, Data or Information Visualization have

become interesting and wide research ﬁelds. If the

main goal of Data Visualization is to communicate

information clearly and efﬁciently to users, involv-

ing the creation and study of the visual representation

of data – i.e., “information that has been abstracted

in some schematic form, including attributes or vari-

ables for the units of information” (Friendly and De-

nis, 2001) – the Information Visualization main task

is the study of (interactive) visual representations

of abstract data to reinforce human cognition. The

abstract data may include both numerical and non-

numerical data, such as text and geographic infor-

mation. The ﬁeld of information visualization has

emerged from research in human-computer interac-

tion, computer science, graphics, visual design, psy-

chology, and business methods. It is increasingly ap-

plied as a critical component in scientiﬁc research,

digital libraries, data mining, ﬁnancial data analy-

sis, market studies, manufacturing production con-

trol, and drug discovery (Bederson and Shneiderman,

2003). Furthermore, the challenges that the Big Data

imperative (Caldarola et al., 2015; Caldarola et al.,

2014; Caldarola and Rinaldi, 2015) imposes to data

management severely impact on data visualization.

The “bigness” of large data sets and their complexity

in term of heterogeneity contribute to complicate the

representationof data, making the drawing algorithms

complex. To make an example, let us consider the

popular social network Facebook, in which the nodes

represent people and the links represent interpersonal

connections; we note that nodes may be accompanied

by information such as age, gender, and identity, and

links may also have different types, such as colleague

relationships, classmate relationships, and family re-

lationships. The effective representation of all the in-

formation at the same time is challenging. The most

common solution is to use visual cues, such as color,

shape, or transparency to encode different attributes.

At the same time, the availability of large data coming

from human activities, exploration and experiments,

together with the investigations of new and efﬁciently

ways of visualizing them, open new perspectivesfrom

which to view the world we live in and to make busi-

Caldarola, E., Picariello, A., Rinaldi, A. and Sacco, M.

Exploration and Visualization of Big Graphs - The DBpedia Case Study.

DOI: 10.5220/0006046802570264

In Proceedings of the 8th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2016) - Volume 1: KDIR, pages 257-264

ISBN: 978-989-758-203-5

257

ness. The Infographics become Infonomic, a compos-

ite term between the term Information and Economics

that wield information as a real asset, a real oppor-

tunity to make business and to discover the world.

Various techniques have been proposed for graph vi-

sualization for the last two decades and they will be

presented in the next section. As far as we can say

here, the principled representation methodology we

agree on is the Visual Information Mantra presented

by Keim in (Keim et al., 2008). It can be summarized

as follows: “Analyze ﬁrst, show the important, zoom,

ﬁlter and analyze further, details on demand”.

The reminder of the paper is organized as follows.

After a literature review on graph visualization tech-

niques and methodologies and a review of existing

works speciﬁcally focused in DBpedia exploration

contained in section 2, a description of the case study

and its data ecosystem is provided in section 3. Af-

terward, starting from an explanation of the approach

used for the DBpedia Ontology Schema import proce-

dure within Neo4j, in section 4, the attempts for im-

porting and visualizing the DBpedia extensional part

are described in section 5 and 6. Finally, section 7

draws the conclusion summarizing the major ﬁndings

and outlining future investigations.

2 RELATED WORKS

Since the study conducted in this paper consists in

the visual representation of DBpedia inside the Neo4j

graph DB (Webber, 2012), this section focuses mainly

on a literature review in Graph Visualization, refer-

ring to other well-known works in the literature for a

complete review of the techniques and theories in In-

formation Visualization (Spence, 2001; Mazza, 2009;

Fayyad et al., 2002; Ware, 2012). Graphs are tra-

ditional and powerful tools for visually representing

sets of data and the relations among them by draw-

ing a dot or circle for every vertex, and an arc be-

tween two vertices if they are connected by an edge.

If the graph is directed, the direction is indicated

by drawing an arrow. Basically, there are generally

accepted aesthetic rules to draw a graph (Purchase,

1997), which include: distribute nodes and edges

evenly, avoid edge crossing, display isomorphic sub-

structures in the same manner, minimize the bends

along the edges. However, since it is quite impossi-

ble to meet all rules at the same time, some of them

conﬂict with each other or they are very computation-

ally expensive, practical graphical layouts are usu-

ally the results of compromise among the aesthetics.

There exists different graph visualization layouts in

the literature, such as: the Tree Layout, the Space

Division Layout, the Matrix Layout and the Spring

Layout (Fruchterman and Reingold, 1991), to men-

tion a few. The latter will be used in this work and

it worth to spend few words on it. The Spring Lay-

out, also known as Force-Directed layout, is a popular

strategy for general graph visualization. The strategy

consists in modeling the graph as physical systems

of rings or springs. The attractive idea about spring

layout is that the physical analogy can be very natu-

rally extended to include additional aesthetic informa-

tion by adjusting the forces between nodes. As one of

the ﬁrst few practical algorithms for drawing general

graphs, spring layout is proposed by Eades in 1984

(Eades, 1984). Since then, his method is revisited

and improved in different ways (Gansner and North,

1998). Mathematically, Spring layout is based on a

cost (energy) function, which maps different layouts

of the same graph to different non-negative numbers.

Through approachingthe minimum energy, the layout

results reaches better and better aesthetically pleasing

results. The main differencesbetween differentspring

approaches are in the choice of energy functions and

the methods for their minimization.

Speciﬁcally regarding the visualization of DBpe-

dia, there are not many works in the literature. The

most part of them confront with the LOD (Linked

Open Data) paradigm where DBpedia represents

the nucleus of the whole ecosystem. In (Brunetti

et al., 2012) a formal Linked Data Visualization

Model (LDVM) has proposed to dynamically con-

nect data with visualizations, by incorporating ana-

lytical extraction and visual abstraction steps, while

in (Helmich et al., 2014) the authors use the LDVM

implemented by Payola (a framework for analysis and

visualization of Linked Data) to visualize RDF Data

Cubes. On the visualization of RDF (Resource De-

sciption Framework) data in general, (Cammarano

et al., 2007) proposes an automatic technique for

mapping data attributes to visualization attributes for-

mulating this problem as a schema matching prob-

lem. In (Chan et al., 2009) a Web-based visualiza-

tion system has been proposed in order to incorporate

data integration into an iterative, interactive data ex-

ploration and analysis process of Wikipedia over DB-

pedia semantic graph. Finally, in (Bikakis and Sel-

lis, 2016) the authors survey the systems developed

by the Semantic Web community in the context of

the Web of Linked Data (WoD), and discuss to which

extent these satisfy the contemporary requirements,

summing up that WoD community should consider

scalability and performance as vital requirements for

the development of the future exploration and visu-

alization systems, but also should provide more so-

phisticated mechanisms that capture users preferences

KDIR 2016 - 8th International Conference on Knowledge Discovery and Information Retrieval

258

and assist them throughout large data exploration and

analysis tasks. By adopting this last advice, our work

aim at easing the exploration of large data like DBpe-

dia, by exploiting the features of new tools emerged

in the Big Data landscape.

3 DBpedia CASE STUDY

The case study presented in this paper consists in

the partial reiﬁcation of the DBpedia dataset inside

the Neo4J graph database. This one is an ACID-

compliant transactional database with native graph

storage and processing where everything is stored in

form of either an edge, a node or an attribute (Web-

ber, 2012; Robinson et al., 2013; Vukotic et al.,

2015). DBpedia is a crowd-sourced community ef-

fort to extract structured information from Wikipedia

and Wikidata and make this information available on

the Web (Lehmann et al., 2015). By extracting the

structured content of Wikipedia, such as infoboxes,

tables, lists, and categorization data, and putting them

in a consistent knowledge base, it is possible to an-

swer expressive queries otherwise not allowed via the

text search engine inside Wikipedia: Give me all cities

in New Jersey with more than 10,000 inhabitants or

Give me all Italian musicians from the 18th century,

are just few examples. More importantly, with the ad-

vent of the Web of Data, several data providers have

started to publish and interlink data on the Web, in-

spired by the Tim Berners-Lees Linked Data princi-

ples and since DBpedia consists in a generalist knowl-

edge base covering different domains such as mu-

sic, science, sports, arts, and so forth, various data

providers have started to set RDF links from their data

sets to DBpedia, making the latter one of the central

interlinking-hubs of the emerging Web of Data.

DBpedia is a very large dataset. The current

English version (2015-04) describes 4.58 million

”things” with 583 million ”facts”. In addition, DB-

pedia presents localized versions in 125 languages,

which describe 38.3 million things. Altogether the

DBpedia 2015-04 release consists of 3 billion pieces

of information (RDF triples) out of which 583 million

were extracted from the English edition of Wikipedia,

2.46 billion were extracted from other language edi-

tions, and about 50 million are links to external data

sets. 4.22 million resources are classiﬁed in a consis-

tent ontology, including 1,445,000 persons, 735,000

places (including 478,000 populated places), 411,000

creative works (including 123,000 music albums,

87,000 ﬁlms and 19,000 video games), 241,000 or-

ganizations (including 58,000 companies and 49,000

educational institutions), 251,000 species and 6,000

diseases. All this information are stored inside Vir-

tuoso RDF Store and are available for download in

different formats. Mainly, DBpedia project provides

ﬁles as N-Triples and N-Quads, where the N-Quads

version contains additional provenance information

for each statement, but a tabular version of some of

the core DBpedia data sets are provided as CSV and

JSON ﬁles. In the following subsections, a detailed

description of the DBpedia data will be outlined.

3.1 DBpedia Schema and Dataset

The DBpedia data ecosystem consists of two funda-

mental parts: the DBpedia Schema (i.e., the insten-

sional part, the DBpedia Ontology) and the proper

dataset (i.e., the extensional part). The current ver-

sion of the Schema contains 735 classes, 1,098 ob-

ject properties, 1,583 datatype properties, 132 spe-

cialized datatype properties, 408 owl:equivalentClass

and 200 owl:equivalentProperty mappings external

vocabularies. The extensional part of DBpedia con-

tains the canonicalized and the localized datasets.

While the ﬁrst ones contain triples extracted from

the respective Wikipedia whose subject and object

resource have an equivalent English article, the sec-

ond ones contain also the triples whose URIs do

not have an equivalent English article. A URI is a

Uniform Resource Identiﬁer (URI) and consists in

a string of characters used to identify a resource.

Its international extension (IRI) may contain charac-

ters from the Universal Character Set (Unicode/ISO

10646), including Chinese or Japanese kanji, Ko-

rean, Cyrillic characters, and so forth, and allowing

identiﬁcation of resources in other languages. All

DBpedia IRIs/URIs in the canonicalized datasets use

the generic namespace http://dbpedia.org/resource/.

The localized datasets use DBpedia IRIs (not

URIs) and language-speciﬁc namespaces, e.g.,

http://pt.dbpedia.org/resource/Brasilia.

Each thing in the DBpedia data set is de-

noted by a de-referenceable IRI- or URI-based ref-

erence of the form http://dbpedia.org/resource/Name,

where Name is derived from the URL of the

source Wikipedia article, which has the form

http://en.wikipedia.org/wiki/Name, so that each DB-

pedia entity is tied directly to a Wikipedia article. Ev-

ery DBpedia entity name resolves to a description-

oriented Web document (or Web resource). Each DB-

pedia entity is described by various properties, the

most important being: the label, a short and long En-

glish abstract, a link to the corresponding Wikipedia

page, and a link to an image depicting the thing (if

available).

In this study, we focus on the canonicalized

Exploration and Visualization of Big Graphs - The DBpedia Case Study

259

Table 1: DBpedia data set ﬁles.

Dataset Description

Mapping-based

Types

Contains the rdf:types of the instances which have been extracted from the infoboxes

Mapping-based

Properties

Contains the actual data values that have been extracted from infoboxes (e.g., ’vol-

ume’).

Mapping-based

Properties (Speciﬁc)

Contains properties which have been specialized for a speciﬁc class using a speciﬁc

unit. e.g. the property height is specialized on the class Person using the unit centime-

tres instead of metres.

Titles Contains the rdfs:labels representing the title of the corresponding Wikipedia page

Short Abstract Contains the short abstract for each DBpedia resource. It is the value of the

rdfs:commentproperty

Extended Abstracts Contains the long abstract for each DBpedia resource. It is the value of the dbpe-

dia.org/ontology/abstract property.

External Links Contains the wikipedia External links. For example, the books.google.com URL for a

DBpedia Book resource.

dataset, and in particular on the English version of

dataset ﬁles. Table 1 reports a not comprehensive list

of ﬁles available from the download section of DBpe-

dia.org website, with a short description of their con-

tent. They are the most generic ones (the most im-

portant ones) and those we need in our work. All ﬁles

are available in N-Triples (.nt and .nq) and Turtle (.ttl)

format.

4 IMPORTING AND

VISUALIZING THE DBpedia

ONTOLOGY IN Neo4J

The ﬁrst step in creating and visualizing the DBpe-

dia data set inside a running instance of Neo4J Graph

database has regarded its intensional schema, i.e.,

the DBpedia Ontology Schema. In order to accom-

plish this step, the Jena API and the Neo4J Java API

have been used for accessing the schema ontology

serialized in a owl ﬁle and, later on, for converting

its classes and properties in nodes and edges inside

Neo4J. A simple mapping mechanism has been used:

each ontological class has been converted in a node

with an automatically generated id and three proper-

ties: class URI, class label and class comment. Subse-

quently, the class nodes have been linked together via

two types of edges: LINK and SUBCLASSOF. The

ﬁrst refers to the ontological object properties link-

ing two classes each other, while the second refers to

the hierarchical is-a relationship existing between two

classes in the ontology. The ﬁrst type of edge has got

an automatically generated id, the object property la-

bel and the object property URI, while the second has

no properties. All this information are retrieved from

the DBpedia ontology owl ﬁle. Figure 1 shows a large

scale representation of the DBpedia ontology. For the

sake of clarity and to avoid a confused cloud of words,

the ﬁgure shows only a bunch of nodes with their la-

bels and an arc between two nodes if a object property

exists between the corresponding classes in the on-

tology. The ﬁgure shows only the largest connected

subgraph of the DBpedia Ontology not the whole on-

tology. Conversely, ﬁgures 2 and 3 show an excerpt of

the DBpedia Schema converted into a Neo4J labelled

graph and visualized in Cytoscape with a simple cus-

tom style. The ﬁrst image highlights some of the link

relations existing between nodes correspondingto ob-

jected properties in the DBpedia schema and uses a

force-directed layout to draw the graph, while the sec-

ond one shows the hierarchical relationships subclas-

sof existing between classes and uses a Sugiyama-like

algorithm (Sugiyama, 2002) to draw the hierachical

links existing between nodes.

5 IMPORTING AND

VISUALIZING THE DBpedia

DATA SET

For the extensional part of DBpedia graph, we have

focused on two data set ﬁles, namely, the Mapping-

based Types and the Mapping-based Properties, leav-

ing out the speciﬁc mapping-based properties ﬁle be-

cause it contains specialized properties which go be-

yond the scope of this work. In fact, the procedure

described here is a generalistic one for importing and

visualizing DBpedia resources, regardless the speciﬁc

features of their ontological nature. Before digging

into the procedure details, it worths to have a close

look at the ﬁrst two ﬁles. The Mapping-based Types

ﬁle contains rdf triples in the form: subject rdf:type

KDIR 2016 - 8th International Conference on Knowledge Discovery and Information Retrieval

260

Figure 1: A large scale view of the DBpedia Ontology

Schema.

Figure 2: First view of the DBpedia Schema in Neo4J: the

object properties.

object. Each triple describes an instance (i.e., a DB-

pedia resource) of a particular type, e.g., a Person, a

Book, or just a Thing. An excerpt of the ﬁle follows:

<h ttp : / / dbpedia . org / r eso ur ce / Autism> <h tt p : / /www.w3 . org

/1999/02/22 − rdf −syntax −ns # type> <h tt p : / / dbpedia . org /

ontolog y / Disease> .

<h ttp : / / dbpedia . org / r eso ur ce / Anarchism> <h t t p : / /www. w3. org

/1999/02/22 − rdf −syntax −ns # type> <h tt p : / /www. w3. org

/2 002 /07 /owl#Thing> .

<h ttp : / / dbpedia . org / r eso ur ce / Alabama> <h tt p : / /www. w3. org

/1999/02/22 − rdf −syntax −ns # type> <h tt p : / / dbpedia . org /

ontolog y / Admin i s t r ati v e R e g i on> .

<h ttp : / / dbpedia . org / r eso ur ce / Abraham

Lincoln> <h tt p : / /www.

w3. org /1999/02/22 − rdf −syntax −ns #type> <h tt p : / / dbpedia

. org / on tol o g y / Of fi c e H older> .

On the other hand, the Mapping-based Properties

ﬁle contains rdf triples in the form: subject property

object. An excerpt of the ﬁle follows:

<h ttp : / / dbpedia . org / r eso ur ce / Alf r ed

Hitchcock> <h tt p : / /

dbpedia . org / ontolog y / almaMater> <h tt p : / / dbpedia . org /

r e so ur ce / Enfield

Town> .

<h ttp : / / dbpedia . org / r eso ur ce / Amitabh

Bachchan> <h tt p : / /

dbpedia . org / ontolog y / almaMater> <h tt p : / / dbpedia . org /

r e so ur ce / Na i ni t al> .

<h ttp : / / dbpedia . org / r eso ur ce / Aloi s

Alzheimer> <h ttp : / /

dbpedia . org / ontolog y / i n s t i t u t i on > < h tt p : / / dbpedia . org

/ r e so ur ce / F r ankf u r t> .

<h ttp : / / dbpedia . org / r eso ur ce / Alber t o

Giacomett i > <h tt p : / /

dbpedia . org / ontol o g y / t ra in i ng > <h tt p : / / dbpedia . org /

r e so ur ce / Geneva> .

<h ttp : / / dbpedia . org / r eso ur ce /A.

E . Housman> <h ttp : / / dbpedia

. org / on t o logy / almaMater> <h t tp : / / dbpedia . org / r eso u rc e

/ Oxford> .

<h ttp : / / dbpedia . org / r eso ur ce / British

Army> <h tt p : / / dbpedia .

org / ontology / country> < ht tp : / / dbpedia . org / r e sou rce /

El iz ab et h

I I > .

5.1 DBpedia Mapping-based Types

All DBpedia instances have been imported in Neo4J

by mapping them to speciﬁc Neo4J nodes with two

properties: the id (automatically generated) and the

URI, in the form: http://dbpedia.org/resource/Name,

where Name is the label of the corresponding in-

stance. All nodes are linked to other nodes through

oriented arcs as resulting in the mapping-based types

ﬁle. Technically, it has been created a ﬁrst map con-

taining the node id and the DBpedia resource URI and

a second map (serialized in a CSV ﬁle) containing a

list of pairs in the form (subject id, object id) that rep-

resents the subject and the object of the rdf:type prop-

erty.

Figure 4 shows the results of the following query

in Cypher:

match (n {URI: ’http://dbpedia.org/ontology/Book’})

-[r: rdf_type]-(m)

return n,r,m limit 10

Figure 5 shows the results of this other query in

Cypher:

match (n

{URI: ’http://dbpedia.org/resource/Aristotle’})

-[r: rdf_type]-(m)

return n,r,m

This ﬁgure shows all the semantic categories to

which Aristotle belong.

An important difference holds between the two

ﬁgures above. While the ﬁrst one put a DBpedia

ontology class (Book) at the center and links it to a

limited number of instances through incoming arcs

around it, the second one, put a DBpedia resource,

an instance, at the center (Aristotle) and links it to the

semantic categories to which Aristotle belong. In this

case, the arcs from the center nodes outcome.

5.2 DBpedia Mapping-based Properties

The import of DBpedia mapping-based properties has

been accomplished through a proceduresimilar to that

described in the previous section, the only difference

being the presence of the extra ﬁeld (predicate) in the

csv ﬁle containing the mappings. These ones have

been serialized in the form: (subject

id, predicate id,

object

id). Figure 6 shows an excerpt of the mapping-

based properties graph inside Neo4J resulting from

Exploration and Visualization of Big Graphs - The DBpedia Case Study

261

Figure 3: Second view of the DBpedia Schema in Neo4J: the hierarchical properties.

Figure 4: The instances of the class Book limited to 10.

Figure 5: All the semantic categories of the resource Aris-

totle.

the following below. This representation is centered

on the resource (Aristotle in this case) and shows all

the properties attached to it.

MATCH (n {URI:

’http://dbpedia.org/resource/Aristotle’})-[r:p]

->(m)

RETURN n,r,m LIMIT 25

6 INCREMENTAL IMPORT AND

VISUALIZATION OF DBpedia

INSTANCES THROUGHOUT

THE ONTOLOGY SCHEMA

The DBpedia.org team provides another important

way to download all the data instances from the exten-

sional dataset, that is, DBpedia as csv tables. Techni-

cally, for each class in the DBpedia ontology (such

as Book, Person, Film, Writer, or Band), DBpedia

provides a single CSV or JSON ﬁle containing all in-

stances of this class. Each instance is described by

its URI, an English label and a short abstract, the

mapping-based infobox data describing the instance

(extracted from the English edition of Wikipedia),

geo-coordinates, and external links. Altogether 685

CSV/JSON ﬁles are provided, packed in a single ﬁle

or in a separate CSV/JSON ﬁle for each class for

download. Each ﬁle starts with 4 headers:

1. The properties labels;

2. The properties URIs;

3. The properties range labels;

4. The properties range URIs.

The ﬁrst column of each ﬁle contains the URIs

of the instances of the current class. The remaining

columns contain the values for the properties that are

deﬁned for this class in the DBpedia ontology. In our

work, we have taken the Book.csv and the Writer.csv

ﬁles containing all book instances and all authors of

written works in DBpedia, respectively, and have tried

to link each instance of Book with the corresponding

instance of Writer. Book.csv contains 67 properties la-

bels with an equal number of properties URIs, range

labels and range URIs in the header section and 31030

books, while Writer.csv contains 109 properties labels

with an equal number of properties URIs, range la-

bels and range URIs and 29995 writers. For this ex-

ploratory study, we have taken into consideration only

the ﬁrst fourth properties of Book beside the instance

URI, namely, the rdf label, the rdf comment, the au-

thor label and the author, and the ﬁrst two properties

of Writer: the rdf label and the rdf comment. The

bridge between the Book class and the Writer class is

the author property of Book, whose URI in the DB-

pedia ontology is http://dbpedia.org/ontology/author

and the range is the set of Person whose URI

KDIR 2016 - 8th International Conference on Knowledge Discovery and Information Retrieval

262

Figure 6: Aristotle properties in DBpedia.

is http://dbpedia.org/ontology/Person. To deal with

these ﬁles, we have written a simple routine in Java

that incrementally import instances from semantic

categories, which are related through speciﬁc link-

ing properties. In this example, the Book category is

linked via the author property to the Writer category,

as explained previously. In this preliminary work,

the liking property has been individuated manually

and only two classes have been imported in the run-

ning instance of Neo4J. Figure 7 shows an excerpt of

the Neo4J graph obtained through the ontology based

incremental import of DBpedia categories described

here. With simple cypher query like the following:

MATCH (b: Book)-[r:author]->(w: Writer)

RETURN b,r,w LIMIT 25

it is possible to obtained bunch of nodes (books)

surrounding the author node (writer), like in the ﬁg-

ure that shows all the books by Agatha Christies. The

use of DBpedia Ontology can automatize as much as

possible the individuation of link properties between

classes. Generally, an ontology-based approach in vi-

sualization of large graphs (Rinaldi, 2008; Cataldo

and Rinaldi, 2010) can improve performance by out-

lining a semantic-based strategy in populating and ex-

ploring labeled properties graphs.

7 CONCLUSIONS

In this paper, we have described an experience of

importing, exploring and visualizing DBpedia into a

graph database. We have confronted with both the

intensional schema and the extensional data set, and,

obviously, the latter has resulted the most challenging.

One consideration immediately arises: the large scale

and exhaustive representation of DBpedia is merely

a chimera. Beyond the scalability and performance

Figure 7: Agatha Christie’s books.

issues, the visualization of large graph data requires

simpliﬁcation and the deﬁnition of a clear style which

put order in the resulting confused and dense structure

of nodes and edges, but also it requires a strategy for

importing and exploring data.

As the ﬁrst attempt to load DBpedia instances

has gone in the direction of importing all the types

instances and all the mapping-based generic proper-

ties from DBpedia ﬁles, an intricate net of nodes has

rapidly exploded in graph visualizer tools when tried

to navigate from one instance to another, without ﬁl-

tering the query and consequently reducing the num-

ber on nodes involved. For this reason, an incremen-

tal approach in importing and visualizing instances,

throughout the ontology schema, has resulted more

practical. Additionally, a similar approach allows to

extends the knowledge graph by engulﬁng in the orig-

inal nucleus the knowledge domains in the neighbor-

hood of the starting core, in a fashion that is very close

Exploration and Visualization of Big Graphs - The DBpedia Case Study

263

to the way in which the human mind organizes con-

cepts and links them with each other(Rinaldi, 2008).

Moreover the use of efﬁcient techniques to visualize

large knowledge bases could be useful in advanced

ontology based and mutimedia applications (Rinaldi,

2009; Moscato et al., 2010; Rinaldi, 2014). How-

ever, it presents a major drawback: the need to deﬁne

manually the linking properties between instances be-

longing to two or more ontology classes.

Nevertheless, this approach worths future investi-

gations and research works, mainly in the attempt to

automatize as much as possible the choice of proper-

ties that link classes each other.

REFERENCES

Bederson, B. B. and Shneiderman, B. (2003). The craft

of information visualization: readings and reﬂections.

Morgan Kaufmann.

Bikakis, N. and Sellis, T. (2016). Exploration and visual-

ization in the web of big linked data: A survey of the

state of the art. arXiv preprint arXiv:1601.08059.

Brunetti, J. M., Auer, S., and Garc´ıa, R. (2012). The linked

data visualization model. In International Semantic

Web Conference (Posters & Demos).

Caldarola, E. G., Picariello, A., and Castelluccia, D. (2015).

Modern enterprises in the bubble: Why big data mat-

ters. ACM SIGSOFT Software Engineering Notes,

40(1):1–4.

Caldarola, E. G. and Rinaldi, A. M. (2015). Big data:

A survey - the new paradigms, methodologies and

tools. In Proceedings of 4th International Conference

on Data Management Technologies and Applications,

pages 362–370.

Caldarola, E. G., Sacco, M., and Terkaj, W. (2014). Big

data: The current wave front of the tsunami. ACS Ap-

plied Computer Science, 10(4):7–18.

Cammarano, M., Dong, X., Chan, B., Klingner, J., Talbot,

J., Halevy, A., and Hanrahan, P. (2007). Visualization

of heterogeneous data. Visualization and Computer

Graphics, IEEE Transactions on, 13(6):1200–1207.

Cataldo, A. and Rinaldi, A. M. (2010). An ontological ap-

proach to represent knowledge in territorial planning

science. Computers, Environment and Urban Systems,

34(2):117–132.

Chan, B., Talbot, J., Wu, L., Sakunkoo, N., Cammarano,

M., and Hanrahan, P. (2009). Vispedia: on-demand

data integration for interactive visualization and ex-

ploration. In Proceedings of the 2009 ACM SIG-

MOD International Conference on Management of

data, pages 1139–1142. ACM.

Eades, P. (1984). A heuristics for graph drawing. Congres-

sus numerantium, 42:146–160.

Fayyad, U. M., Wierse, A., and Grinstein, G. G. (2002). In-

formation visualization in data mining and knowledge

discovery. Morgan Kaufmann.

Friendly, M. and Denis, D. J. (2001). Milestones

in the history of thematic cartography, statis-

tical graphics, and data visualization. URL

http://www.datavis.ca/milestones.

Fruchterman, T. M. and Reingold, E. M. (1991). Graph

drawing by force-directed placement. Softw., Pract.

Exper., 21(11):1129–1164.

Gansner, E. R. and North, S. C. (1998). Improved force-

directed layouts. In Graph Drawing, pages 364–373.

Springer.

Helmich, J., Kl´ımek, J., and Neˇcask`y, M. (2014). Visualiz-

ing rdf data cubes using the linked data visualization

model. In The Semantic Web: ESWC 2014 Satellite

Events, pages 368–373. Springer.

Keim, D., Andrienko, G., Fekete, J.-D., G¨org, C., Kohlham-

mer, J., and Melanc¸on, G. (2008). Visual analytics:

Deﬁnition, process, and challenges. In Information

visualization, pages 154–175. Springer.

Lehmann, J., Isele, R., Jakob, M., Jentzsch, A., Kon-

tokostas, D., Mendes, P. N., Hellmann, S., Morsey,

M., van Kleef, P., Auer, S., et al. (2015). Dbpedia–

a large-scale, multilingual knowledge base extracted

from wikipedia. Semantic Web, 6(2):167–195.

Mazza, R. (2009). Introduction to information visualiza-

tion. Springer Science & Business Media.

Moscato, V., Picariello, A., and Rinaldi, A. M. (2010). A

recommendation strategy based on user behavior in

digital ecosystems. In Proceedings of the Interna-

tional Conference on Management of Emergent Digi-

tal EcoSystems, pages 25–32. ACM.

Purchase, H. (1997). Which aesthetic has the greatest effect

on human understanding? In Graph Drawing, pages

248–261. Springer.

Rinaldi, A. M. (2008). A content-based approach for doc-

ument representation and retrieval. In Proceedings of

the eighth ACM symposium on Document engineer-

ing, pages 106–109. ACM.

Rinaldi, A. M. (2009). An ontology-driven approach for se-

mantic information retrieval on the web. ACM Trans-

actions on Internet Technology (TOIT), 9(3):10.

Rinaldi, A. M. (2014). A multimedia ontology model based

on linguistic properties and audio-visual features. In-

formation Sciences, 277:234–246.

Robinson, I., Webber, J., and Eifrem, E. (2013). Graph

databases. ” O’Reilly Media, Inc.”.

Spence, R. (2001). Information visualization, volume 1.

Springer.

Sugiyama, K. (2002). Graph drawing and applications for

software and knowledge engineers, volume 11. World

Scientiﬁc.

Vukotic, A., Watt, N., Abedrabbo, T., Fox, D., and Partner,

J. (2015). Neo4j in Action. Manning.

Ware, C. (2012). Information visualization: perception for

design. Elsevier.

Webber, J. (2012). A programmatic introduction to neo4j.

In Proceedings of the 3rd annual conference on Sys-

tems, Programming, and Applications: Software for

Humanity, pages 217–218. ACM.

KDIR 2016 - 8th International Conference on Knowledge Discovery and Information Retrieval

264