Consolidating User-defined Concepts with StYLiD

Consolidating User-Consolidating User-defined Concepts with defined Concepts with StYLiDStYLiD

Aman Shakya1, Hideaki Takeda1, Vilas Wuwongse2

1National Institute of Informatics Tokyo, Japan

2Asian Institute of Technology, Pathumthani, Thailand

OutlineOutlineIntroduction

◦ Background◦ Social Semantic Web◦ Problems

Proposed approach◦ Overview◦ The StYLiD system◦ Concept Consolidation◦ Application Scenarios

Related WorkConclusion and Future Work

4 Feb. 2009 ASWC 2008, Bangkok, Thailand 2

BackgroundBackgroundPeople share data on the Web

Unstructured data

Structured DataModel different types of “things”Concepts, schemas – attributes and

relationsPossible with Semantic Web technologiesAdvantages

Semantic applications, automation, integration, interoperability, effective search and browsing

3ASWC 2008, Bangkok, Thailand4 Feb. 2009

ChallengesChallengesLong Tail of information domains (Hunyh et al.

2007)◦ Wide variety of data to share

Not enough Ontologies

Ontology creation is a difficult process◦ Not feasible for every new type of data

Ontologies are difficult to understand and use◦ Semantic Web tech. too complex for ordinary people


Social Semantic WebSocial Semantic WebSocial Software

◦ Easy to understand and use◦ Incremental & dynamic publishing platforms◦ Mass participation◦ Social interaction and collaboration

Semantic Web◦ Structured Data, automation / interoperability ,

etc.

Social software + Semantic Web◦ Social Semantic Web◦ Collaborative knowledge creation and sharing


6

Collaborative Knowledge Creation

Collaborative Knowledge Base

Users Users

4 Feb. 2009 ASWC 2008, Bangkok, Thailand

7

Problems1. Creation of data models satisfying many

people and contexts simultaneously

2. Existence of Multiple Conceptualizations◦ Different user requirements, perspectives or

contexts

◦ But information exchange/integration should be possible

3. Consensus by collaborative interaction may be difficult and time-consuming

4. Still difficult for ordinary people◦ Considerable learning curve for existing systems

◦ Restrictive constraints


8

Knowledge Sharing by Loose Collaboration

Collaborative Knowledge Base

Users

Users

Local KB

Local KB

Local KB

Users


ObjectivesObjectives1. To enable ordinary people to share a wide

variety of structured data on the Semantic Web.

2. To allow multiple conceptualizations of the same concept by different people.

3. Consolidation of multiple user-defined concept schemas to form collaborative definitions.

4. To facilitate the emergence of informal lightweight ontologies.


10

Overview

Social Platformfor

Structured Data Authoring

Concept Grouping

External Resources

Concepts

Instances

Structured Data Collection

Browsing, Searching,Services

Concept groups

Concept Consolidation

Schema Alignment

Structured Linked Data Grouped

concepts

User Community

Consolidated Concepts

Emerging Lightweight Ontologies

StYLiDStYLiDStructure Your own Linked Data

http://www.stylid.org (get your account!)

Social Software for Sharing a wide variety of Structured Data

Users can freely define their own concepts Easy for ordinary people

◦ Flexible and relaxed interface for data entry

Consolidate Multiple Concept Schemas◦ To create rich concept definitions

Emerging informal ontologies◦ Popular concepts and evolving definitions


Creating a new Concept

Attribute labels

Description

Suggested Value Range

“Project” concept


Enter Instance Data

Literal value

Suggested range concepts

Resource URI

Multiple Values


Concept ConsolidationConcept Consolidation

Hotel - ver.1 (user1)

Name

Address

Country


Name

Address

Phone-number


Name

Location

Rating


Name

Capacity

Zip-code


Name

Zip-code

Price


Name

Lat

Long

Hotel (user1)

Hotel (user2)

Hotel (user3)

Hotel

14ASWC 2008, Bangkok, Thailand

Virtual Concept

4 Feb. 2009

Allow multiple local conceptualizationsAspects (Takeda et al., 1995), DDL (Borgida and Serafini, 2003), Contextual ontologies, C-OWL (Bouquet et al., 2004), -connections (Kutz et al., 2004 ; Grau et al., 2004)



A concept consolidation C is defined as a triple

< , S, A> where◦ - consolidated concept

◦ S - set of constituent concepts {C1,C2 ,…..Cn}

◦ A is the attribute alignment between and S

Based on Global-as-View (GAV) approach for data integration◦ Global schema defined as views on source schemas

Consolidated Concept with consolidated attributes◦ aligned to source concept attributes as views

CC

C

C


16

C1a2a

ma

iCaligned( , )

aligned( , )

1a 1ia2ia

inia

1ia

2a 2ia

aligned( , )ma inia

)( 1ia)( 2

ia

)( inia

view

1C

nC

iM

nM

1M

A = { , … }1M 2M nM

image

< , S, A>C


Concept ConsolidationConcept ConsolidationQuery Unfolding (Advantage of GAV over LAV)

◦Queries over

to queries over {C1,C2 ,…..Cn}

◦Using alignment A◦Union of results

Translation of instances◦From one conceptualization to

anotherTranslation of queries


C

Concept CloudConcept Cloud

Sub-Cloud

Consolidated concept

4 Feb. 2009 18ASWC 2008, Bangkok, Thailand

Alignment of Concept Alignment of Concept SchemasSchemasAttribute Alignments suggested Automatically

◦ Alignment API implementation with WordNet extension

Users verify and complete the alignment◦ Human intelligence + Machine intelligence

Alignments are represented and saved (for everyone)

◦ Alignment ontology (Hughes and Ashpole, 2004)

◦ Alignment API alignment specification language (Euzenat et al., 2007) Other formats : C-OWL, SWRL, OWL axioms, XSLT, SEKT-ML and

SKOS.

Incremental alignment

A Unified View◦ Consolidated concept with Consolidated Attributes◦ Homogenous table of data


Concept versions

x


Search on Consolidated Concept

SPARQL21ASWC 2008, Bangkok, Thailand

Structured SearchStructured Search

4 Feb. 2009

Grouping Similar ConceptsSuggest groups of similar concepts

◦ Under a similarity threshold

ConceptSim(C1, C2) = w1*NameSim(N1, N2) + w2*SchemaSim(S1, S2)

◦ NameSim - WordNet-based similarity (Lin’s algorithm)

◦ SchemaSim - Average similarity of best matching pairs of attributes

◦ Hungarian Algorithm - find best matching pairs (Kuhn,

1955; Munkres, 1957)

Browse groups of similar conceptsVisualize clusters of related concepts


Visualization of Concepts Visualization of Concepts GroupingGrouping

12/10/2008 ASWC 2008, Bangkok, Thailand 23

Cytoscape

Grouping & Consolidation for Grouping & Consolidation for Concept GeneralizationConcept Generalization

Consolidation of related concepts as generalization◦( Hotel + Apartment ) =>

Accomodation

Multiple groupings possible with same concepts◦( Hotel + Restaurant ) => Eating

place

04/21/2312/10/2008 ASWC 2008, Bangkok, Thailand 24

Linked DataLinked DataLink data instances

◦ Select instance URIs as attribute valueLink to external data resources

◦ Enter external URIs as attribute value

Link to Wikipedia contents

ASWC 2008, Bangkok, Thailand 25

StYLiStYLiDD

Wikipedia URI DBpedia URI

4 Feb. 2009

(User friendly) (Machine friendly)

Application ScenariosSocial Site for

Structured Information Sharing

Concept Schemas

Structured data

External Data

Resources

StYLiD

CMS

Data IntegrationSchema

Alignment

Information Sharing Social

Semantic Website

Users

Usershttp://www.stylid.org

Application ScenariosIntegrated Semantic portal


Structured data

External Data

Resources

StYLiD

Data Backend

Data IntegrationSchema

Alignment

Integrated Semantic

Portal

UsersAdmin

Concept Schemas

IS1

IS2

IS3

Wrapper1

Wrapper2

Wrapper3

Information Sources

Related WorkRelated Work

Semantic Blogging

Semantic Wikis

Ontology from Folksonomy◦ Specia & Motta, 2007; Van Damme et al., 2007; Mika, 2007

Schema alignment & Data integration

ASWC 2008, Bangkok, Thailand 284 Feb. 2009

Exhibit

ConclusionConclusion


Consolidation of multiple concepts◦ Allow multiple conceptualizations◦ Loose collaborative approach for concept

definition◦ Data Integration for the Semantic Web

StYLiD◦ Social software for sharing wide variety of

structured Semantic Web data◦ Easy for ordinary users to contribute freely

Emergent lightweight informal ontologies◦ Evolution, Consolidation and Grouping◦ Ontology as by-product of information sharing

4 Feb. 2009

Future WorkFuture Work


Computing hierarchical / non-hierarchical relations among concepts

Better schema alignment techniquesConsolidation of instancesUsing / Mapping to existing

vocabulariesMash-ups / plugins to utilize structured

dataSharing scrapers to collect data from

the web…

4 Feb. 2009

Thank You!Thank You!


QuestionsComments

4 Feb. 2009

Consolidating User-defined Concepts with StYLiD

Documents

wide variety of data

linked data http

knowledge sharing

different people

ordinary peopleflexible

collaborative interaction

userdefined concepts

timeconsumingstill difficult