Ontology Engineering to Enrich Linked Data

IASLOD 2012 -International Asian Summer
School on Linked Data
13-17 Aug. 2012, KAIST, Daejeon, Korea

Ontology Engineering to
Enrich Linked Data
Kouji Kozaki
The Institute of Scientific and Industrial Research (I.S.I.R),
Osaka University, Japan

2012/08/15 IASLOD 2012 1

Self introduction: Kouji KOZAKI
 Brief biography
 2002 Received Ph.D. from Graduate School of Engineering, Osaka University.
 2002- Assistant Professor, 2008- Associate Professor in ISIR, Osaka University.
 Specialty
 Ontological Engineering

 Main research topics
 Fundamental theories of ontological engineering
 Ontology development tool based on the ontological theories
 Ontology development in several domains and ontology-based application
 Hozo(法造) -an environment for ontology building/using- (1996- )
 A software to support ontology（=法） building（=造） and
use
 It’s available at http://www.hozo.jp as a free software
 Registered Users：3,500 (June 2012) Cooperator:
Enegate Co, ltd.
 Java API for application development is provided.
 Support formats: Original format, RDF(S), OWL.
 Linked Data publishing support is coming soon.
2012/08/15 IASLOD 2012 2

My history on Ontology Building
 2002-2007 Nano technology ontology
 Supported by NEDO(New Energy and Industrial Technology Development Organization)
 2006- Clinical Medical ontology
 Supported by Ministry of Health, Labour and Welfare, Japan
 Cooperated with: Graduate School of Medicine, The University of Tokyo.
 2007-2009 Sustainable Science onology
 Cooperated with: Research Institute for Sustainability Science （RISS）, Osaka
University.
 2007-2010 IBMD(Integrated Bio Medical Database)
 Supported by MEXT through "Integrated Database Project".
 Cooperated with: Tokyo Medical and Dental University, Graduate School of Medicine, Osaka U.
 2008-2012 Protein Experiment Protocol ontology
 Cooperated with: Institute for Protein Research, Osaka University.
 2008-2010 Bio Fuel ontology
 Supported by the Ministry of Environment, Japan.
 2009- Disaster Risk ontology
 Cooperated with: NIED (National Research Institute for Earth Science and Disaster Prevention)
 2012- Bio mimetic ontology
 Supported by JSPS KAKENHI Grant-in-Aid for Scientific Research on
Innovative Areas
2012/08/15 IASLOD 2012 3

Agenda
 (1) Trends of Linked Data in Semantic Web
Conferences from ontological viewpoints.

 (2) How ontologies are used in Linked Data
 An analysis of Semantic Web applications.
 9 types of ontology usages x 5 types of ontologies

 (3) Ontology Engineering to Enrich Linked Data

2012/08/15 IASLOD 2012 4

Semantic Web Conference
 ISWC：International Semantic Web Conference
 2001 Symposium@ Stanford University, California, USA
 Participants 245, submissions 58, acceptance rate 60％

 No workshops, 3 tutorials

 2002- Annual conference, Venue: Europe → USA → Asia
 2011 ISWC2011@Bonn, Germany
 Participants 597, submissions 264, acceptance rate 19％

 16 workshops, 6 tutorials

 ESWC：European Semantic Web Conference
 2004 Symposium, 2005- Annual conference.
 2010- Extended Semantic Web Conference.
 ASWC：Asian Semantic Web Conference
 2006- twice / three years
 2011 JIST2011（The Join International Semantic Technology Conference)
 Jointed with CSWC2011 (The 5th Chinese Semantic Web Conference)

2012/08/15 IASLOD 2012 5

Venues of International
Semantic Web Conferences
ISWC ESWC ASWC
SWWS＠California, USA
ISWC2002＠Sardinia, Italy
ISWC2003＠Sanibel Island,FL,USA Symposium@Osaka, WS@Nara

ISWC2004＠Hiroshima, Japan ESWS＠Heraklion, Greece
ISWC2005＠Galway, Ireland ESWC2005＠Heraklion, Greece
ISWC2006＠Athens, GA, USA ESWC2006＠Budva,Montenegro ASWC2006@Beijing,China

ISWC2007&ASWC2007＠Busan,Korea ESWC2007＠Innsbruck, Austria

ISWC2008＠Karlsruhe, Germany ESWC2008＠Tenerife, Spain ASWC2008@Bangkok, Thailand

ISWC2009＠Washington D.C.Area,USA ESWC2009＠Heraklion, Greece ASWC2009@Shanghai, China

ISWC2010＠Shanghai, China ESWC2010＠Heraklion, Greece

ISWC2011＠Bonn.Germany ESWC2011＠Heraklion, Greece JIST2011@Hangzhou, China

ISWC2012＠Boston, USA ESWC2012＠Heraklion, Greece JIST2012@Nara, Japan

ISWC2013＠Sydney, Australia ESWC2013@Montpellier, France (JIST2013@Korea)

2012/08/15 IASLOD 2012 6

JIST 2012, 2-4 Dec. 2012, Nara, Japan
- Submission due : 24 Aug. 2012.
- It has a Special Track on Linked Data
http://www.ei.sanken.osaka-u.ac.jp/jist2012/
2012/08/15 IASLOD 2012 7

Research Trends in
Semantic Web Conferences(1/3)
ISWC ESWC ASWC
ISWC2003＠Sanibel Island,FL,USA Symposium@Osamka, WS@Nara


Basic technologies of Semantic WebGreece mainly discussed.
ISWC2005＠Galway, Ireland ESWC2005＠Heraklion,
are
DAML, OIL→ predecessor of OWL, Rule-ML, Ontology…


Frequency QuestionESWC2009＠Heraklion, Greece
ISWC2009＠Washington D.C.Area,USA
/ Discussion: ASWC2009@Shanghai, China
“I can understand the basic idea of Semantic Web.

However, who describes meta data?”

ISWC2012＠Boston, USA ESWC2012＠Heraklion, Greece

ISWC2013＠Sydney, Australia

2012/08/15 IASLOD 2012 8

Research Trends in
ISWC ESWC ASWC

ISWC2005＠Galway, Ireland ESWC2005＠Heraklion, Greece

As an answer to the question “Who describes ASWC2008@Bangkok, Thailand
ISWC2008＠Karlsruhe, Germany ESWC2008＠Tenerife, Spain meta data?”
Usage of Social Network System, Web2.0 were actively China
ISWC2009＠Washington D.C.Area,USA ESWC2009＠Heraklion, Greece ASWC2009@Shanghai,
FOAF, WiKi …
ISWC2010＠Shanghai,Blog, RSS, ESWC2010＠Heraklion, Greece
discussed. China
・Collaborative Development of Ontologies was one of
hot topics.Australia
ISWC2013＠Sydney,
・Many Semantic Web based applications were developed.
2012/08/15 IASLOD 2012 9

Research Trends in
ISWC ESWC ASWC
★The first presentation ofESWS＠Heraklion, Greece
ISWC2004＠Hiroshima, Japan DBPedia.
(DBPedia was presented also at ESWC2005＠Heraklion, Greece
WWW2007.)
ISWC2005＠Galway, Ireland
A Special Session
ISWC2006＠Athens, GA, USA on Linked Data
ESWC2006＠Budva,Montenegro ASWC2006@Beijing,China


8 3
ISWC2009＠Washington D.C.Area,USA ESWC2009＠Heraklion, Greece ASWC2009@Shanghai, China
10 4 Debate
- Linked Data: Now what?
After DBPedia, Linked Data became the hottest

research topic in Semantic Web Conference.
ISWC2013＠Sydney, Australia

：the numbers of research track papers whose title includes “Linked
Data”.
2012/08/15 IASLOD 2012 10

Summary of the trends in SWC
 Changes of main research topics
 Semantic processing using metadata based on ontologies
 “Who describes meta data?” → Collaborative building, Web2.0
 Linking between Data (instances)：Linked Data
(Ideal) Semantic Web
Rich semantics

× Linked Data
SNS・Web2.0

Simple/ easy to use Tag（RSS,FOAF）

Scalability
2012/08/15 IASLOD 2012 11

ISWC2011/ESWC2011: Keynote
 Keynotes in ISWC2011/ESWC2011 also discussed
trends of Semantic Web research．
 ISWC2011: Keynote by Frank van Harmelen
 10 Years of Semantic Web:
does it work in theory?
Available at http://www.cs.vu.nl/~frankh/spool/ISWC2011Keynote/

 ESWC2011: Keynote by James A. Hendler
 “Why the Semantic Web
will Never Work”
Available at http://www.eswc2009.org/

 Common claims
 Ontology << Data (instance)＝LOD
 LOD is main application in resent Semantic Web

2012/08/15 IASLOD 2012 12

From ISWC2011: Keynote
by Frank van Harmelen
Terminological knowledge
is much smaller than the
factual knowledge

2012/08/15 IASLOD 2012 13

From ESWC2011: Keynote
by James A. Hendler

2012/08/15 IASLOD 2012 14

What does
“Ontology << Data” means?
 It is true that the number of data (instances) linked in LOD is many
more than the number of concepts (types) .
 However, it is not the right claim ”We do not need ontology.”, “Minimum
ontologies are enough (for LOD).” , “Linking data is more important.”．
 Because we can use huge scales of LOD, it is required to deal with their
semantics appropriately and to realize advanced semantic processing.
How to deal
(Ideal) Semantic Web
Rich semantics

with semantics.
It is an important
problem to

× bridge the GAP.
Linked Data How to
use LOD.
SNS・Web2.0

Simple/ easy to use Tag（RSS,FOAF）

2012/08/15 IASLOD 2012 Scalability 15

From ISWC2011：Opening

Not change increase decrease
increase

2012/08/15 IASLOD 2012 16

ISWC2011:Research
Papers
 Research Tracks (three papers in each sessions)
 Web of Data
 Social Web
 User Interaction
 RDF Query - Alternative Approaches How to use
 RDF Query - Performance Issues
Linked Data
 RDF Query - Multiple Sources
 RDF Data Analysis
 Policies and Trust
 MANCHustifications and Provenance
 KR – Reasoners
 KR - Semantics
 Formal Ontology & Patterns How to deal with
 Ontology Evaluation Semantics
 Ontology Matching, Mapping

2012/08/15 IASLOD 2012 17

ISWC2011:Wrokshops
 Consuming Linked Data※
 Detection, Representation, and Exploitation of Events
 Knowledge Evolution and Ontology Dynamics
 Linked Science※
 Multilingual Semantic Web ※Workshops whose main topic
 Ontologies come of Age is Liked Data
 Ontology Matching
 Ordering and Reasoning
 Scalable Semantic Web Knowledge Base Systems
 Semantic Personalized Informaton Management
 Semantic Sensor Networks
 Semantic Web Enabled Software Engineering
 Social Data on the Web
 Terra Cognita - Foundations, Technologies and Applications of the
Geospatial Web
 Uncertainty Reasoning for the Semantic Web
 Web Scale Knowledge Extraction
2012/08/15 IASLOD 2012 18

ISWC2011:Wrokshops
 Consuming Linked Data※
 Detection, Representation, and Exploitation of Events
 Knowledge Evolution and Ontology Dynamics
nd workshop on
 Linked 2
The Science※
 Multilingual Semantic Web Data ※Workshops whose main topic
Consuming Linked
 Ontologies come(participants: 70-80)
・big workshop of Age is Liked Data
 Ontology Matching about 50%
・acceptance rate:
 ・Papers about basic technologies are more than applications.
Ordering and Reasoning
 ★Some organizers (participants) argue Systems
Scalable Semantic Web Knowledge Basethat

“I want to got more Informaton application of
Semantic Personalizedpaper about Management LOD.”

“We have to know (practical/concrete) Needs for LOD”
Semantic Sensor Networks
 Semantic Web Enabled Software Engineering

Linked Data-a-thon
Social Data on the Web
 Terracontest whose theme is to develop LOD application within 2 weeks.
・A Cognita - Foundations, Technologies and Applications of the
・Given Resources for the subject is conference information of ISWC.
Geospatial Web
 Uncertainty Reasoning (All the Semantic Web
・Only 3 submissions. for of them got prize…)
 Web Scale Knowledge Extraction
2012/08/15 IASLOD 2012 19

Agenda
 SW → Web2.0 → LOD
 How to use LOD? How to deal with semantics?
 It is based on my presentation in ASWC2008,
“Understanding Semantic Web Applications”.
 An analysis of Semantic Web applications (including LOD).
 Method: 9 types of ontology usages x 5 types of ontologies


2012/08/15 IASLOD 2012 20

Motivation for
SW application analysis
 Background
 About 10 years after the birth of Semantic Web (SW)
 [A roadmap to the Semantic Web, Sep 1998, Tim Berners-Lee]
 Fundamental technologies for SW
 RDF(S), OWL, SPARQL, SWRL … etc.
 So many SW applications
 In spite of so many efforts on research and development of
SW technologies, “Killer Application” of SW is still
unknown [Alani 05, Motta 06].
 Motivation
 It would be beneficial for us to get an overview of the
current state of SW applications to consider next direction
of SW.
 Our approach
 We analyzes SW Apps from the view point of ontology.
 Especially we focus on “What type of ontologies is used”
and “How ontologies are used.”
2012/08/15 IASLOD 2012 21

Steps for Analyzing SW
Applications from Ontological
Viewpoint
 We analyzed 190 SW applications which utilize
ontologies extracted from Semantic Web conferences
according to the following steps:
 (1) Giving short explanations about the application.
(One sentence for each)
 (2) Identifying the type of usage of ontology
(9 categories).
 (3) Identifying the target domain.
 (4) Identifying types of ontology (5 categories).
 (5) Identifying the language for description.
(RDF(S), OWL, DAML+OIL, …etc)
 (6) Identifying the scale of ontology.
(number of concepts and/or instance models)

 On the way of this analysis, we discussed about the criteria
for classification of applications interactively.
2012/08/15 IASLOD 2012 22

applications which is
analyzed
Number
Conferences Dates Venues
of Apps
International Semantic Web Conference (ISWC)
ISWC2002 Jun. 9-12, 2002 Sardinia, Italy 9
ISWC2003 Oct.20-23, 2003 Sanibel Island,FL,USA 19
ISWC2004 Nov. 7-11, 2004 Hiroshima, Japan 18
ISWC2005 Nov. 6-10, 2005 Galway, Ireland 25
ISWC2006 Nov.5-9, 2006 Athens, GA, USA 26
ISWC2007&ASWC2007 Nov.11- 15, 2007 Busan, Korea 18
European Semantic Web Conference (ESWC)
ESWC2005 May29-Jun.1,2005 Heraklion, Greece 24
ESWC2006 Jun.11-14, 2006 Budva, Montenegro 11
ESWC2007 Jun. 03 - 07, 2007 Innsbruck, Austria 17
Asian Semantic Web Conference (ASWC)
ASWC2006 Sep.3- 7, 2006 Beijing, China 23
※SW and ontology engineering tools (e.g. ontology editors, ontology
alignment tool) are not the target of the analysis.
2012/08/15 IASLOD 2012 23

Steps for Analyzing SW
Applications from Ontological
Viewpoint
 We analyzed 190 SW applications which utilize
ontologies extracted from Semantic Web conferences
according to the following steps:
 (1) Giving short explanations about the application.
(One sentence for each)
 (2) Identifying the type of usage of ontology
(9 categories).
 (3) Identifying the target domain.
 (4) Identifying types of ontology (5 categories).
 (5) Identifying the language for description.
(RDF(S), OWL, DAML+OIL, …etc)
 (6) Identifying the scale of ontology.
(number of concepts and/or instance models)

 On the way of this analysis, the authors discussed about the
criteria for classification of applications interactively.
2012/08/15 IASLOD 2012 24

Types of Usage of Ontology for
a SW Application(1/5)
Types of Usage of Ontology
  Ontology applications scenarios
[Uschold 99]
Shallow (1) Common Vocabulary 1)neutral authoring
 (2) Semantic Search 2)common access to information
 (3) Systematized Index 3)indexing for search
LOD

 (4) Data Schema
 The role of an ontology
1)a common vocabulary[Mizoguchi03]
 (5) Media for Knowledge
2)data structure
Sharing
3)explication of what is left implicit
 (6) Semantic Analysis
4)semantic interoperability
 (7) Information Extraction
5)explication of design rationale
 (8) Rule Set for Knowledge
6)systematization of knowledge
Models 7)meta-model function
Deep  (9) Systematizing Knowledge 8)theory of content
 Basically, a SW application is categorized to one of the types according
to its main purpose.
 Some SW applications which use ontology for multiple ways are
categorized to multiple categories.
2012/08/15 IASLOD 2012 25

 (1) Usage as a Common Vocabulary
 To enhance interoperability of knowledge content, this type of
application uses ontology as a common vocabulary.
 (2)Usage for Search
 This type of application uses semantic information of Index
ontologies for semantic search.
O O ntol
ntol ogy
ogy Us
 (3) Usage as an Index Search
hie
str
 Applications of this category utilize in
not only the index vocabulary defined Ind
Annotation
in ontologies but also its structural
of knowle
information (e.g., an index term’s Common Vocabulary concepts
position in the hierarchical structure) ontology
Usage
as systematized indexes when vocab
accessing the knowledge resources. searc
 e.g.) Indexes for Knowledge Portal, D ocum ents / Law D ata analy
Semantic Navigation D ocum ents / Law D ata
2012/08/15 IASLOD 2012 26

 (4) Usage as a Data Schema
 Applications of this category use ontologies as a data schema to
specify data structures and values for target databases.
 (5) Usage as a Media for Knowledge Sharing
 Applications of this category aim at knowledge sharing among
different systems and/or people using ontologies and instance.
 e. g. knowledge alignment, knowledge mapping, communication
support Reference ontology Ontology A Ontology B

Mapping to the
Reference Ontology Ontology Mapping

Knowledge Knowledge Knowledge Knowledge
A B A B
(i) Knowledge Sharing through (ii) Knowledge Sharing using
a Reference Ontology Multiple Ontologies
2012/08/15 IASLOD 2012 27

 (6) Usage for a Semantic Analysis
 Reasoning and semantic processing on the basis of ontological
technologies enable us to analyze contents which are annotated
by metadata.
 e.g. automatic classification, statistical analysis, validation

 (7) Usage for Information Extraction
 Applications which aim at extracting meaningful information
from the search result are categorized here.
 e.g. Recommendation, extracting some features from web pages ,
summarization of contents

 Comparison among categories (2) Search, (6) and (7):
 (2) Search -> just output search results without modifications.

 (6) Semantic Analysis -> add some analysis to the output of (2)

 (7) Information Extraction -> extract meaningful information
before outputting for users.

2012/08/15 IASLOD 2012 28

 (8) Usage as a Rule Set (Meta Model) for Knowledge Models
 We can use ontologies as meta-models which rule the knowledge
(instance) models.
 Relations between the ontologies and the instance models
correspond to that of the database and the database schema of
category (4).
 Compared to the category (4), Knowledge models need more
flexible descriptions in terms of meaning of the contents.
O ntol
ogy
 (9) Usage for Systematizing Knowledge
 To integrate these usages from (1) to (8),
Meta
Model
ontologies can be used for Knowledge
Systematization.
 e.g. integrated knowledge systems,

knowledge management systems
and contents management systems

D atabases / K now l
edge M odel
s
2012/08/15 IASLOD 2012 29

Types of Ontology
 Characteristics of ontologies
 Design concept
 Focusing on efficient information processing

 Focusing on good conceptualizations to capture the

target world accurately as much as possible
 Semantic feature Without depending on other characteristics
 cf. An ontology spectrum [Lassila and McGuninness 01]
 Target domains
 Building process (How to be constructed)
 By hand, by machine learning, by collaborative work

 Description languages
 The scale of ontology
 Number of concepts and instances, Scalability, Coverage

2012/08/15 IASLOD 2012 30

Types of Ontology
 5 Categories from the viewpoint of semantic
feature of ontologies.
LOD
 (A) Simple Schema
 e.g. RSS and FOAF for uniform description of data for SW.

RDF(S)
OWL
OWL SWRL
 (B) Hierarchies of is-a Relationships among Concepts
 A light-weight ontology described by Only rdfs:subClassOf.
e.g. Hierarchies of topics on Web portal, controlled Vocabulary.

＋
 (C) Relationships other than “is-a” is Included
 Other various relationships (properties) with some
Restriction (e.g. cardinality, all/someValuesFrom).
 (D) Axioms on Semantics are Included
 Specifying further constraints among the concepts or instance
by introducing axioms on semantic constraints (e.g. “transitive
Property”, “inverseOf”, “disjointWith” , “one of” ).
 (E) Strong Axioms with Rule Descriptions are Included
 Further description of constraints on the category (D) with rule
descriptions (e.g. KIF or SWRL).
2012/08/15 IASLOD 2012 31

Results of the Analysis

The result of our analysis is available at the URL:
http://www.hozo.jp/OntoApps/
2012/08/15 IASLOD 2012 32

Distribution of Types of Usage of
Ontology
イプの分布 Mainly deal with
There is not so big difference among
利用タof usage.
the ratios of each type “data” processing
1)共通語彙 Vocabulary
(1) Common
4% 4%
(2) Search
2)検索
20% 19% 3)イIndex ス
(3) ンデック
LOD
4)データ Schema
(4) Data
スキーマ
(5) Knowledge Sharing
5)知識共有の媒体
8% 11% (6) Semantic Analysis
6)分析
(7) Information Extraction
7)抽出
9%
13% (8) Knowledge Modeling
8)知識モデルの規約
12%
9)知識の体系化 Systematization
(9) Knowledge

Most of current studies in the SW Explicitly deal with
application deal with “data” “knowledge” processing
processing on structured data.
2012/08/15 IASLOD 2012 33

Distribution of Types of
Ontology
A few ontologies have
Rule descriptions.
オント (A) Simple Schema
ロジーの種類の分布
(E) Strong Axioms with Rule
Descriptions are Included 3% 1% (B) Hierarchies of is-a
(D) Axioms on Semantics Relationships
6%
are Included 11% among 簡易スキーマ half of the
Almost
Concepts systems use OWL
概念階層 extended OWL.
or
(C) Other Relationships その他の関係
Unknown,
are Inculuded 意味制約12%
79% Others,
DAML 公理あり
12% OWL,
+OIL,
4%
OWL-S,
Most of the SW applications use
ontologies including a variety
50%
RDF(S),
types of relations. 23%

2012/08/15 IASLOD 2012 34

A Correlation between the Types of
Usage and the Types of Ontology
The Types of O ntol
ogy
m e (B ) Is-a (C ) O ther
(A ) Si pl (E) Rul
e
(D )A xi s
om Total
erarchi Rel onshi
Schem a H i es ati p D escri ons
pti
s
(1) C om m on V ocabulary 0 4 7 0 0 11
(2) Search 1 2 43 4 1 51
(3) Index 0 3 23 3 0 29
(4) D ata Schem a 0 0 32 5 0 37
(5) Know ledge Shari ng 1 0 31 1 0 33
(6) Sem anti A nal s
c ysi 1 1 21 3 0 26
(7) Inform ati Extracti
on on 1 2 15 3 0 21
(8) Know ledge M odelng
i 0 1 36 9 8 54
(9) Know ledge System ati on
zati 0 2 8 1 0 11
Total 4 15 216 29 9 273

2012/08/15 IASLOD 2012 35

A Correlation between the Types of
Usage and the Types of Ontology
The Types of O ntol
ogy
m e (B ) Is-a (C ) O ther
(A ) Si pl (E) Rul
e
(D )A xi s
om Total
erarchi Rel onshi
Schem a H i es ati p D escri ons
pti
s
(1) C om m on V ocabulary 0 4 7 0 0 11
(2) Search 1 2 43 4 1 51
(3) Index
(4) D ata Schem a
0
0
3
0
LOD23
32
3
5
0 29
0 37
(5) Know ledge Shari ng 1 0 31 1 0 33
(6) Sem anti A nal s
c ysi 1 1 21 3 0 26
(7) Inform ati Extracti
on on 1 2 Semantic3Web 0 21
15
(8) Know ledge M odelng
i 0 1 36 9 8 54
(9) Know ledge System ati on
zati 0 2 8 1 0 11
Total 4 15 216 29 9 273
Deeper type of usage needs deeper used in mainly
Rule description is
semantic feature of ontologies. modeling.
knowledge
2012/08/15 IASLOD 2012 36

Conference Transition of the
Types of Usage
会議毎の利用タイプの推移

The amount of papers surveyed in each conference
40
9 19 18 24 25 11 23 26 17 18 (9) Knowledge
The amounts of types of usage

(9) Knowledge Sys
35 Systematization
(7)
(8) Knowledge Mo
(8) Knowledge
30 Modeling
(6) (7) Information Ex
25 (7) Information
Extraction Analy
(6) Semantic
20 (5) (6) Semantic
(5) Knowledge Sha
Analysis
15 (4) (5) Knowledge
(4) Data Schema
Sharing
(3) Index
10
(4) Data Schema
5 (2) (3) Index
(2) Search
(2) Search
(1) Common Vocab
0 (1) Common
Vocabulary

2012/08/15 IASLOD 2012 37

Conference Transition of the
Types of Usage application development focuses on
The mainstream of SW
data processing, and overcoming the difficulty of knowledge
会議毎の利用タイプの推移

processing might paperskey to create conference About 20
The amount of be a surveyed in each killer applications.
40
9 19 18 24 25 11 23 26 17 18 (9) Knowledge
The amounts of types of usage

The amounts higher-level semantic
the use for of types of usage are (9) Knowledge Sys
35 processing ((4)-(9)) are increasing Systematization
increasing year by year. (7)
(8) Knowledge Mo
(8) Knowledge
30 gradually. Modeling
(6) (7) Information Ex
25 (7) Information
Extraction Analy
(6) Semantic
20 (5) (6) Semantic
(5) Knowledge Sha
Analysis
15 (4) (5) Knowledge
(4) Data Schema
Sharing
(3) Index
10
(4) Data Schema
5 (2) (3) Index
(2) Search
(2) Search
(1) Common Vocab
0 (1) Common
there is no significant change in the use of ontology Vocabulary
as vocabulary or for retrieval ((1)-(3))

2012/08/15 IASLOD 2012 38

The Combinations of the
Types of Usage
(1) Vocabulary (2) Search
利用タイプの分布
(3) Index

1)共通語彙 Vocabulary
(1) Common
4% 4%
(2) Search
2)検索
20% 19% (3) Index ク
3)インデッス
(4)Data Schema (5) Knowledge (6) Semantic (4) Data Schema
4)データスキーマ
Sharing Analysis
6)分析
(7) Information Extraction
7)抽出
9%
(7) Information
Extraction 12% (9) Systematization
(8) Knowledge
Modeling
Knowledge
(9) Knowledge
9)知識の体系化
Systematization

2012/08/15 IASLOD 2012 39

The Combinations of the
Types of Usage
(1) Vocabulary
(7)
(2) Search
(3) Index

(2) 1)共通語彙 Vocabulary
(1) Common
(6) 4% 4%
(2) Search
2)検索
20% 19% (3) Index ク
3)インデッス
(4)Data Schema (5) Knowledge (6) Semantic (4) Data Schema
Sharing
(2) Search, (6)Analysis and of (2) search and
The combinations Analysis
(5) Knowledge sharing
(7)Info. Extraction are (5) Knowledge Sharing
->integrated search across several
usages mainly for semantic
6)分析
retrieval. information resources.
->(1) common vocabularies (7) Information Extraction
7)抽出
tend to be used for search 9%
systems. 13% (8) Knowledge Modeling
(7) Information (8) Knowledge (9) Knowledge
Extraction Modeling12% Systematization (9) Knowledge
Combined with all other 9)知識の体系化
Combined with (8) Knowledge
types systematically. Systematization
modeling more frequently
in compare with (2) Search and
(6) Semantic Analysis.
2012/08/15 IASLOD 2012 40

The distribution of the types
of usage per a domain(1/2) イプ
ドイン毎の利用タ
メ
Domains
(number of systems) The number of the types of usage Multipurpose
multipurpose(27) (1) Common Vo
multimedia(24) Multimedia
service(21)
access management(3)
利用タイプの分布 Service
(2) Search
(3) Index
software(9) Software 1)共通語彙 Vocabulary
(1) Common (4) Data Schema
ontology(7) 4% 4%
(2) Search
2)検索 (5) Knowledge S
agent(2)
Webpage(11) Webpage
19% (3) Index ク (6) Semantic An
3)インデッス
20%
Wiki(4) (7) Information
Web community(6) (4) Data Schema
knowledge (8) Knowledge M
Semantic Desktop(4)
management (5) Knowledge Sharing
Knowledge Management(9) …
knowledge (9) Knowledge S
business(17)
6)分析
e-government(4) Business (7) Information Extraction
7)抽出
geographical(4)
9% Scientific information
education(4) 13% (8) Knowledge Modeling
scientific information(13) 12%
bio(9) Bio (9) Knowledge
medical(11) Medical Systematization

2012/08/15 0 10 IASLOD 2012
20 30 40 41
50

Types of U sage of O ntol
ogy
The distribution and servicetypes the percentage
In the software of the domains,
1) 2) 3) 4) 5) 6) 7) 8) 9)

of KM and ✓domain(2/2) percentage of (9)
In
per aontology domains, the
of usage (8) knowledge modeling isishigher in comparison
✓
knowledge systematization higher.
✓
with scientific domains
✓
✓
✓ ✓ ✓ ✓ 1)共通語彙 Vocabulary
(1) Common
✓ ✓ 4% 4% The numbers of the Search
✓ ✓ 2)検索 for
(2) use
higher-level semantic
✓ (3) Index ス
20%
✓ processing ((4)-(9)) are ク
19% 3)インデッ
✓ increasing gradually.Data Schema
✓ (4)
✓
✓
8% ✓ ✓ ✓ 11% (6) Semantic Analysis
6)分析
✓
✓ (7) Information Extraction
7)抽出
9%
✓
✓ ✓
12%
✓ (9) Knowledge
✓ Systematization
✓ ✓
✓
✓ ✓ ✓ ✓
✓ ✓ scientific domains
2012/08/15 ✓ ✓
IASLOD 2012 42

Summary:
analysis of SW applications
 Summary
 Analysis of 190 SW applications from the viewpoint of
 Types of Usage of Ontology for a SW Application
 Types of Ontology .
 This classifications can be applied to LOD apps.
 The result of our analysis is available at the URL:
 http://www.hozo.jp/OntoApps/

 Open questions
 How rich semantics are needed for LOD?
 It is important viewpoints of the users (domain expert).
 Ontology can add richer semantics to LOD, but is it
valuable to pay building cost?
 We have to consider balance between cost and benefit.
2012/08/15 IASLOD 2012 43

Agenda

 An analysis of Semantic Web applications.
 9 types of ontology usages x 5 types of ontologies


2012/08/15 IASLOD 2012 44

Ontology Engineering to
Enrich Linked Data
 Features of ontology in class level
 It reflects understanding of the target world.
 Well organized ontologies have generalized rich knowledge
based on consistent semantics.
 Ontologies are systematized knowledge of domains.
 My research interest on LOD
 How can I use ontologies in class level for semantic processing?
 When I combine it with LOD, how does it enrich LOD?
 Possible applications
 Flexible viewpoint management from multi-perspectives.
 Integrated understanding support of domain experts.
 Idea/Innovation supporting system.

2012/08/15 IASLOD 2012 45

Examples
 Understanding an Ontology through
Divergent Exploration
 Presented at ESWC2011
 Ontology of disease
 “River Flow Model of Diseases”
 presented at ICBO (International Conference on Biomedical
Ontology) 2011
 Dynamic Is-a Hierarchy Generation System
based on User's Viewpoint
 Presented at JIST2011

2012/08/15 IASLOD 2012 46

Motivation: Understanding an
Ontology through Divergent
Exploration
 Issue: A serious gap exists between interests of
ontologists and domain experts
 Ontologists try to cover wide areas domain-independently
 Domain experts are well-focused and interest in domain specificity.
→Ontologies are sometimes regarded as verbose and too general by
domain experts
Understanding the target
Interest in common
world from the domain- GAP properties of concepts
specific viewpoints
and generality.

Experts in policy Target World
×
Ontologists
Motivation:ecosystem
Experts in It is highly desirable to have
Ontology
Knowledge
Knowledge knowledge structuring from the general perspective
not only
sharing × the domain-specific and multiple-perspectives.
systematization
isbut also from
difficult
Experts in energy
2012/08/15 IASLOD 2012 47

Divergent exploration of
ontology
It bridges the gap between
ontologies and domain experts
Understanding
Capturing of the essential
from the domain- GAP conceptual structure
specific viewpoints
②On the fly reorganizing
as generally as possiblesome
conceptual structures from the
Experts in policy Target World ontology as visualizations
×
Ontology developer Conceptual
Experts in ecosystem map
Ontology
①Systematizing the
× Experts in policy
conceptual in energy
Experts
structure focusing
on common characteristics ✓
Knowledge sharing
is difficult
Experts in energy Experts in ecosystem
✓
It would stimulate their Integrated understanding of
intellectual interests and could the ontology and cross-
support idea creation domain knowledge
2012/08/15 IASLOD 2012 48

(Divergent)
Ontology exploration tool
1) Exploration of multi-perspective conceptual chains
2) Visualizations of conceptual chains
Visualizations as
Exploration of an ontology conceptual maps from
different view points

“Hozo” – Ontology Editor
Multi-perspective conceptual chains
represent the explorer’s understanding of
ontology from the specific viewpoint. Conceptual maps
2012/08/15 IASLOD 2012 49

Node represents Is-a (sub-class-of)
a concept relationshp Referring to
(=rdfs:Class) another concept

slot represents
a relationship
(=rdf:Property)

2012/08/15 IASLOD 2012 50

Viewpoints for exploration
■The viewpoint as the combination of a starting point and an aspect.
・The aspect is the manner in which the user explores the ontology.
It can be represented by a set of methods for tracing concepts according
to its relations.
Aspects for tracing concept
Starting point
rdfs:subClassOf
Related relationships
Kinds of extraction
in Hozo in OWL
(1) Extraction of sub concepts
Aspects (A) is-a relationship rdfs:subClassOf
(2) Extraction of super concepts
Extraction of concepts referring to other
properties which (3)
(B)
part-of/attribute-of
are referred in
concepts
relationship
owl:restriction
(4) Extraction of concepts to be referred to
Depending on (5) Extraction of contexts
(C)
Other properties
relationship (6) Extraction of role concepts
play(playing) (7) Extraction of player (class constraint)
(D) relationship (8) Extraction of role concepts
+ restriction on property names
and/or tracing classes
2012/08/15 IASLOD 2012 51

System architecture
A Java client application version and
a web service version are available.
Ontology Exploration Tool Browsing conceptual
maps using web browser
Ontology exportation

Publish conceptual
conceptual
aspect dialog map visualizer maps on the Web
Connections with
Connections with
Connections with
other web
other web
other web
Concept tracing module
concept extraction module systems through
systems through
systems through
concepts defined
concepts defined
concepts defined
in the ontology
in the ontology
in the ontology
import
Hozo-ontology editor OWL ontology
Legends
Ontology building inputs by users flows of data
commands
2012/08/15 IASLOD 2012 52

Option settings for
exploration

Selected relationships
Kinds of aspects
are traced and shown as
links in conceptual map

property
names

constriction
tracing classes
Conceptual map visualizer Aspect dialog
2012/08/15 IASLOD 2012 54

Explore the focused
(selected) path.

2012/08/15 IASLOD 2012 55

Search
Path Ending point (1)
Selecting of ending points
Finding all possible
paths from stating
point to ending points

Starting point

Ending point (2)
Ending point (3)

2012/08/15 IASLOD 2012 56

Search
Path

Selected ending points

2012/08/15 IASLOD 2012 57

Functions for ontology
exploration
 Exploration using the aspect dialog:
 Divergent exploration from one concept using the aspect dialog
for each step
 Search path:
 Exploration of paths from stating point and ending points.
 The tool allows users to post-hoc editing for extracting only
interesting portions of the map.

 Change view:
 The tool has a function to highlight specified paths of conceptual
chains on the generated map according to given viewpoints.
 Comparison of maps:
 The system can compare generated maps and show the common
conceptual chains both of the maps.
2012/08/15 IASLOD 2012 58

Usage and evaluation of
ontology exploration tool
 Step 1: Usage for knowledge structuring in
sustainability science

 Step 2: Verification of exploring the abilities of the

 Step 3: Experiments for evaluating the ontology
exploration tool

2012/08/15 IASLOD 2012 59

structuring in sustainability
science
 Sustainability Science (SS)
 We aimed at establishing a new interdisciplinary
scheme that serves as a basis for constructing a
vision that will lead global society to a
sustainable one.
 It is required an integrated understanding of the
entire field instead of domain-wise knowledge
structuring.
 Sustainability science ontology
 Developed in collaboration with domain expert in
Osaka University Research Institute for
Sustainability Science (RISS).
 Number of concepts：649, Number of slots： Sustainability Science
1,075 http://en.ir3s.u-tokyo.ac.jp/about_sus
 Usage of the ontology exploration tool
 It was confirmed that the exploration was fun for
them and the tool had a certain utility for
achieving knowledge structuring in sustainability RISS, Osaka Univ.
science. [Kumazawa 2009]

2012/08/15 IASLOD 2012 60

Verification of exploring capability of
If we ask domain experts to explore the SS ontology using the tool and
verify whether it can generate maps they wish to do, it means that we
verify not only exploring capability of the ontology exploration tool but
also the ontology itself.
 Verification method
1) Enrichment of SS ontology
The enriched the SS ontology on the basis of 29 typical scenarios which a domain
We concepts appearing in these
expert organized problem structures in biofuel domains by reviewing existing research.
scenarios were extracted and
generalized to add into scenario reproducing operations
2) Verification of the ontology
We verified whether the ontology exploration tool could generate conceptual maps
which represent original scenarios.
burn agriculture=（deforestation, soil deterioration caused by farmland development for
Result
biofuel crops）⇒ harvest sugarcanes （air pollution caused by intentional burn），disruption of

ecosystem93% (27/29) of original scenarios were successfully reproduced as

caused by deforestation（water pollution）
conceptual maps.
 The rest (2 scenarios) could not be reproduced because we missed to
Example: Air pollution, cause of forest fire, soil deterioration, water pollution are attributed
add some relationships in the ontology.
to intentional burn when forest is logged or sugarcanes are harvested in the
We can conclude that the for biofuel crops. ability of the tool is sufficient.
farmland development exploration

2012/08/15 IASLOD 2012 61

Usage and evaluation of
 Step 1: Usage for knowledge structuring in
sustainability science

 Step 2: Verification of exploring the abilities of the

 Step 3: Experiments for evaluating the ontology
exploration tool
 1) Whether meaningful maps for domain experts were obtained.
 2) Whether meaningful maps other than anticipated maps were
obtained.
Maps which are representing the contents of the scenarios anticipated
by ontology developers at the time of ontology construction.
Note: the subjects don’t know what scenarios are anticipated.
2012/08/15 IASLOD 2012 62

Experiment for evaluating
 Experimental method
1) The four experts to generated
conceptual maps with the tool in
accordance with condition settings of
given tasks.
2) They remove paths that were
apparently inappropriate from the
paths of conceptual chains included in
the generated maps.
The subjects: 3) They select paths according to their
4 experts in different fields. interests and enter a four-level general
A: Agricultural economics evaluation with free comments.
B: Social science
(stakeholder analysis) A: Interesting
C: Risk analysis B: Important but ordinary
D: Metropolitan environmental
planning
C: Neither good or poor
D: Obviously wrong
2012/08/15 IASLOD 2012 63

Experimental results (1)
Table.2 Experimental results . l
Number of Path distribution based on general evaluation
selected paths A B C D a
Expert A 2 2
Expert A
(second time) 1 1

Expert B 7 4 1 2
Task 1
Expert B
(second time) 6 3 3

Expert C 8 1 5 2
Expert D 3 1 1 1
Expert A 1 1 E
Task 2
Expert B 6 5 1 n
Expert C 7 2 4 1 in
Expert D 5 3 1 1
Expert B 8 4 2 2 c
Task 3 Expert C 4 2 2 n
Expert D 3 3 p
Total 61 30 22 8 1
2012/08/15 IASLOD 2012 64

Table.2 Experimental results . l
Number of maps
Number of Path distribution based on general evaluation
generated: 13 selected paths A B C D a
Expert A 2 2
Number of paths evaluated:1 61
Expert A
1
(second time)
A: Expert B
Interesting 307 (49%) 4 1 85%
2
B: Expert B
Important but6 ordinary 22 (36%)
Task 1
3 3
C: Expert C good or poor 8(13%)5
Neither
(second time)
8 1 2
D: Expert D
Obviously wrong 1(2%)
3 1 1 1
Expert A 1 1 E
We can conclude that the tool could generate
Task 2
Expert B 6 1 5 n
Expert C 7 4 1 2 in
maps or paths sufficiently meaningful for experts.
Expert D 5 1 1 3
c
Expert B 8 4 2 2
n
Number of paths
Task 3 Expert C 4 2 2
Expert D 3 3 p
evaluated: 61
Total 61 30 22 8 1
2012/08/15 IASLOD 2012 65

 Quantitatively comparison of the anticipated maps with the
maps generated by the subjects
(N) Nodes and links (M) Nodes and links included
included in the paths in the paths of generated and
of anticipated maps selected by the experts

50 50 150

About half (50%) of N∩M
the paths About 75% of paths in the
included in the anticipated maps generated maps are new paths
were included in the maps which is not anticipated from
generated by the experts. the typical scenarios .

It is meaningful enough to claim a positive support for the developed tool.
This suggests that the tool has a sufficient possibility of presenting
unexpected contents and stimulating conception by the user.
2012/08/15 IASLOD 2012 66

Exploration of ontology
vs. exploration of linked data
Paths expected by Paths generated by
ontology developers the experts

50 50 150 New paths which is
unexpected from
at the time of
ontology construction.
Paths expected Unexpected (Main) Target
by developer paths of exploration
Exploration of
Liked Data
✓ Instance level

Exploration of
Ontology
✓ ✓ Class level

Liked data is based on a more rich ontologies
→more meaningful paths through divergent.
2012/08/15 IASLOD 2012 67

Summary: Understanding an
Ontology through Divergent
Exploration
 Divergent exploration of an ontology
 It supports to bridge a gap between interests of ontologists and
domain experts and contributes to integrated understanding of an
ontology and its target world from multiple viewpoints.
 Usage and evaluation of the tool
 Usage for knowledge structuring in sustainability science
 Verification of exploring the abilities of the ontology exploration tool
 Experiments for evaluating the ontology exploration tool
 Domain experts could obtain meaningful knowledge for themselves as
conceptual chains through the divergent exploration of the SS ontology.
 Future plans
 Improvements of the tool to support more advanced problems such as
consensus-building, policy-making and so on.
 Application of the ontology exploration tool for ontology refinement.
 An evaluation of the tool on other ontologies (especially in OWL) .
 Divergent exploration of instances (like liked data) with an ontology.

2012/08/15 IASLOD 2012 68

A consensus-building support system
・Display multiple concept
Map
maps
2 ・Highlight common concepts
Map ・Highlight different concepts
1
Map
4
Touch-Table
Map
3

2nd Step: Collaborative workshop

1ｓｔ Step: Individual concept map
2012/08/15 IASLOD 2012
creation 69

The first experimental workshop using
the consensus-building support
system

Discussion using
integrated maps displayed
on a touch-table display

Participants
- 5 experts in sustainability science
- 4 students in environmental engineering
2012/08/15 IASLOD 2012 70

Medical ontology project in Japan
 Developed ontologies
 Disease ontology：
 Definitions of diseases as causal
chains of abnormal state.
 6000+ diseases
 Anatomy ontology：
 Connections between blood vessel,
nerves, bones : 10,000+
 It based on ontological frameworks
(upper level ontology) which can
apply to other domains
 Models for causal chains
 Abnormal state ontology for data
integration
 General framework to define
complicated structures
2012/08/15 IASLOD 2012 71

An example of causal chain
constituted diabetes.
possible causes and effects

… … … …

Type I diabetes …
… Destruction of Diabetes Elevated level Diabetes-related
pancreatic Lack of insulin I
beta cells in the blood
Deficiency
of insulin
of glucose in
the blood
Blindness

loss of sight
… …
Legends
Long-term steroid
treatment … Disorder (nodes)
… Causal Relationship
Steroid diabetes … Core causal chain of a disease
(each color represents a disease)

2012/08/15 IASLOD 2012 72

An example of causal chain
constituted diabetes.
possible causes and effects

… … … …

Type I diabetes …
… Destruction of Diabetes Elevated level Diabetes-related
pancreatic Lack of insulin I
beta cells in the blood
Deficiency
of insulin
of glucose in
the blood
Blindness

loss of sight
… …
Legends
Long-term steroid
Based on abnormal state ontology causal chains defined in
treatment … Disorder (nodes)
…
each areas are generalized and organized across domains.
Causal Relationship
Steroid diabetes … Core causal chain of a disease
(each color represents a disease)
MD in 12 areas describe definitions (causal chains) of disease
2012/08/15 IASLOD 2012 73

Visualizing/reasoning
causal chains in human body

• As the result, we obtained causal
chains which include about 17,000
clinical disorders defined in 6,000
diseases. They represent possible
causal chains in human body.
• We also developed a browsing tool
to visualizes causal chains.
• We also consider publishing the
disease ontology as LOD.
2012/08/15 IASLOD 2012 74

Motivation: Dynamic Is-a Hierarchy
Generation System based on User's
Viewpoint
Understanding
 Domain experts often want to understand the from their own
target world from their own domain-specific viewpoints
viewpoint.
Disease
 In some domains, there are many ways to
categorize the same kinds of concepts.
How diseases are named
 named by the major symptom disease classification by
 diabetes, angina… the symptom
 named by the abnormal object infarction stenosis hyperglucemia
 heart disease, … disease disease disease

 named by the cause of the disease Myocardial
Stroke Angina diabetes
 Myocardial infarction, stroke infarction
 named by the specific environment
 Altitude sickness, … disease classification by the
disease abnormal object
 named by the discoverer
heart brain blood
 Grave’s disease… disease disease disease
Myocardial
infarction
diabetes Stroke Angina
Myocardial
infarction
One is-a hierarchy of diseases cannot
cope with such a diversity of viewpoints. Several is-a hierarchies of diseases
according to their viewpoints
2012/08/15 IASLOD 2012 75

Existing approaches
 Acceptance of multiple ontologies Multiple-inheritance
based on the different perspectives infarction
disease
heart
disease
 Multiple-inheritance, Ontology mapping
Myocardial
Problem infarction

 If we define every possible is-a hierarchy
using multiple-inheritances or ontology Ontology mapping
mapping, they would be very verbose and disease
the user’s viewpoints would become implicit.
infarction stenosis hyperglycemia
disease disease disease

 Exclusion of the multi-perspective Myocardial
infarction Stroke Angina diabetes
nature of domains from ontologies
 The OBO Foundry
disease
 A guideline for ontology development stating
that we should build only one ontology in
heart brain blood
each domain. disease disease disease

Myocardial
infarction

2012/08/15 IASLOD 2012 76

Our approach
Multi-perspective issue Dynamic Is-a Hierarchy
Understanding Generation based on User's
from their own
viewpoints
Viewpoint

Disease

Generation of
is-a hierarchies
We take a user-centric
approach based on
ontological viewpoint
management.
Ontology Viewpoints
Use single-inheritance

2012/08/15 IASLOD 2012 77

Our approach: Dynamic is-a Hierarchy
Generation according to User’s
Viewpoint
classification by
disease the symptom
various is-a hierarchies
infarction stenosis hyperglycemia based on individual perspectives
classification by the
Myocardial
infarction
Stroke Angina diabetes abnormal object
disease

perspective A
「focus on heart brain blood
symptoms」
parts of human body
abnormal state Myocardial
infarction
Angina Stroke diabetes
heart brain blood
infarction stenosis hyperglycemia perspective B
disease 「focus on abnormal
objects」
Myocardial (2) Reorganizing some
diabetes Stroke Angina
infarction conceptual structures from
(1) Fixing the conceptual structure of an the ontology on the fly as
ontology using single-inheritance based visualizations to cope with
on ontological theories various viewpoints.
2012/08/15 IASLOD 2012 78

Our approach: Dynamic is-a Hierarchy
Generation according to User’s
Viewpoint
Multi-perspective issue Dynamic Is-a Hierarchy
Understanding Generation based on User's
from their own
viewpoints
Viewpoint

Disease

Generation of
is-a hierarchies
We take a user-centric
approach based on
ontological viewpoint
Ontology Viewpoints
management.
Use single-inheritance
We propose a framework for dynamic is-a hierarchy generation
according to the interests of the user and implement the framework as an
extended function of “Hozo-our ontology development tool”.
2012/08/15 IASLOD 2012 79

Ontology Engineering to Enrich Linked Data

Ontology Engineering to Enrich Linked Data

Recommandé

Recommandé

Contenu connexe

Similaire à Ontology Engineering to Enrich Linked Data

Similaire à Ontology Engineering to Enrich Linked Data (20)

Plus de Kouji Kozaki

Plus de Kouji Kozaki (20)

Dernier

Dernier (20)

Ontology Engineering to Enrich Linked Data