a preliminary approach on ontologybased visual query formulation for big data
DESCRIPTION
A preliminary approach on ontologybased visual query formulation for big data - MTSR 2013TRANSCRIPT
A preliminary approach on ontology-based visual query formulation for
big data
Ahmet Soylu University of Oslo
MTSR 2013
Thursday, 21 November 2013
Outline
! Introduction
! Background
! Challenges and Requirements
! Optique Approach
! Discussion and Outlook
! The Big Picture
Introduction
IT expert Domain expert
Introduction
Query formulation bottleneck
Introduction
Simple'Case'
'''
Complex'Case'
'''
Op.que'Solu.on'
''
Applica'on*
End-user*
End-user*
End-user*
predefined*queries*
informa'on*need* specialized*query*
Applica'on*
Op'que* ontology-based*queries* Query*Transla'on*
translated*queries*
uniform**sources*
disparate**sources*
disparate**sources*
IT*expert*
informal*
limited*
possible*mismatch*
flexible* op'mised*
Introduction
(Laney, 2001)
Query formulation and query evaluation/answering
Introduction
Simple'Case'
'''
Complex'Case'
'''
Op.que'Solu.on'
''
Applica'on*
End-user*
End-user*
End-user*
predefined*queries*
informa'on*need* specialized*query*
Applica'on*
Op'que* ontology-based*queries* Query*Transla'on*
translated*queries*
uniform**sources*
disparate**sources*
disparate**sources*
IT*expert*
informal*
limited*
possible*mismatch*
flexible* op'mised*
Optique: scalability
Query formulation
Query evaluation (answering)
Background
Visual Query Systems and Languages (Catarci, 1997; Epstein, 1991)
Direct manipulation (Shneiderman, 1983)
Background
! Early approaches: database schema, object-oriented models etc. (e.g., QBE, QBD*, TableTalk etc. )
! Unnatural: flattening & scattering (i.e., normalization/join) ! Ontology-based approaches (e.g., Catarci, 2004; Barzdins, 2009)
! Natural: knowledge representation & reasoning ! Current work suffer from lack of ontology-based data
access (OBDA) frameworks and remain at experimental stages
Background
SQL$REWRITE& REWRITE&
Ontology&(OWL)& mappings&
Q$ QI$
disparate&sources&Visual&Query&System&
End>user&SPARQL$ SPARQL$
QII$
query$transforma5on$
Siemens'(GBs/day)&
Statoil'(GBs/day)&
Energy&Tribunes&
Drilling&FaciliEes&
Expressivity&Usability&
User'
System'
Explore&Construct&
IT&expert&
RDBMS&(TBs)&
Visual Query Systems + Ontology-based Data Access (OBDA) (cf. Rodriguez-Muro, 2012; Kogalovsky, 2012)
Challenges and Requirements
! Two main pillars: ! Expressiveness ! Usability: effectiveness, efficiency, user satisfaction
! Main data access activities: ! Exploration (i.e., understanding the reality of interest) ! Construction (i.e., formulation)
Challenges and Requirements
! Expressiveness: end-user perspective ! What domain constructs to communicate? (e.g., subclass,
disjoint classes etc.) ! What query constructs types to express? (e.g., topological
and non-topological)
! Usability: discern, comprehend, and communicate ! What representation paradigms, interaction styles and visual
attributes? ! How to avoid large and incomprehensible views? ! How to orient user in a large conceptual space? ! How to alleviate Big Data affect?
Optique approach
Optique approach
Optique approach: architecture
! Widget-based mashup: flexible and extensible
Optique approach: design
! A visual query system ! Multi-paradigm
! Diagram, list, form etc. ! Query by Navigation, range selection etc.
! View and Overview ! Faceted search: Amazon, eBay etc.
! data intensive ! hard to join concepts ! good at selection and projection
! Navigational approach: the Web ! hard to do selection and projection ! good at join
Discussion and Outlook
! Expressiveness
! categories of queries, ! 1st level: linear and tree-shaped conjunctive queries ! 2nd level: disjunctive queries, cyclic queries, and aggregation ! 3rd level: negation, aggregation, and universal quantifiers
! A layered/spiral approach ! A VQS is likely to be less expressive than the underlying
formal textual language
Discussion and Outlook
! Usability ! Interactive visualizations
! Gradual and iterative (e.g., node retraction and expansion) ! Collaborative Query formulation and query reuse ! Big Data effect:
! Adaptation and recommendations ! Schema clustering and summarization ! Widgets for context-tailored representations
! Reactive Scenarios
The Big Picture
! Ontology and mapping management ! Time and streams ! Query transformation (incl. optimization) ! Distributed query execution (incl. parallelization)
Applica'on* Query*Formula'on*
Ontology*&*Mapping*Management*
Query*Transforma'on*
Mappings*Ontology*
Query*Planning*
results*
End=user*
IT*
Expe
rt*
query*
Stream*adapter* Query*Execu'on* Query*Execu'on*
.*.*.*Site*B*
.*.*.*Site*C*
*Site*A*
.*.*.*
streaming*data*
Q&A
Thanks!
www.optique-project.eu www.ahmetsoylu.com