#1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany...

20
Panel ADVCOMP/SEMAPRO Luc Vouligny , moderator

Transcript of #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany...

Page 1: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Panel ADVCOMP/SEMAPRO

Luc Vouligny, moderator

Page 2: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Computing Challenges withSemantics and Ontology Models

• Cristovâo D P Sousa– Universidade do Porto, Portugal

• Michel ClauB– Technische Universität, Chemnitz, Germany

• Jason Gu– Dalhousie University, Canada

• Artem Katasonov– VTT, Finland

• Luc Vouligny– Moderator– Hydro-Québec (IREQ)

Page 3: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Cristovâo D P SousaUniversidade do Porto, Portugal

• Semantic artefacts dynamicity.How to reuse conceptual models and how toachieve the trade-off between informal andformal representation.

Page 4: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Michel ClauBTechnische Universität, Chemnitz, Germany

• Challenges for data integration in industrialsettings (aspects: standards, dynamics,performance, robustness, privacy) as they arediscussed along with the 4th industrial revolution.

• Semantic technologies and data quality: differentways to look at data quality, conditions for gooddata quality, sources of impairment,opportunities and limits of semantic technologiesfor better data quality

Page 5: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Jason GuDalhousie University, Canada

• Handling and Modeling Data on SmartTransportation Management Systems.

Page 6: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Artem KatasonovVTT, Finland

• How semantic technology can stay relevant in the BigData age?Can performance ever match the needs?

• Semantic database vs semantic data virtualization(virtual semantic views on non-semantic data)?Our money is on the latter…

Page 7: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Faculty of Mechanical Engineering Department of Factory Planning and Factory Management

1 www.tu-chemnitz.de/mb/FabrPlan Nice, France· July 22, 2015 · Dr.-Ing. Michael Clauß

•  Subtopic 1: Challenges for Data Integration in Industrial Settings •  Aspects: Standards, Dynamics, Performance, Robustness, Privacy

•  Reference to the Discussion of the Fourth Industrial Revolution (Industry 4.0)

Panel on ADVCOMP/SEMAPRO Topic: Computing Challenges with Semantics and Ontology Models

•  Subtopic 2: Semantic Technologies and Data Quality •  Different Ways to look at Data Quality

•  Conditions for good Data Quality and Sources of Impairment

•  Opportunities and Limits of Semantic Technologies for better Data Quality

Page 8: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

How Semantic Technology Can StayRelevant in the Big Data Age?

@ArtemKatasonovVTT Technical Research Center of Finland Ltd

Panel on ADVCOMP/SEMAPRO 2015,Nice, July 22

Page 9: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Click to edit Master title style

Click to edit Master text styles

Second level

Third level

Fourth level

Fifth level

22

Old Question : How to make computers act in a waywe perceive as intelligent?

Three known approaches :

Tradition AI: complex algorithms on raw data, attempt toimitate human logic

Semantic Technology: simpler algorithms on carefullyprepared data (standardized, context-free, unambiguous),still imitating human logic

Big Data (aka Google-way): even simpler algorithms(mostly, nearest-neighbor search) on huge amounts ofexample data, attempt to imitate human memory

Page 10: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Click to edit Master title style

Click to edit Master text styles

Second level

Third level

Fourth level

Fifth level

33

Big Data is clearly winning, and currently appears to bethe only meaningful overall framework for AI.

The other approaches have to be satisfied with asupporting role, at best. Some of Traditional AIalgorithms (feature extraction, optimizations, etc.) areroutinely used to further improve Big Data –basedsystems.

What is the role of Semantic Technology in the Big Dataage?

Actually, in principle, opportunities are clear:

Mapping of data models of questions and of example data.

Connecting various heterogeneous example data sets.

Allowing questions at various levels of abstraction.

Page 11: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Click to edit Master title style

Click to edit Master text styles

Second level

Third level

Fourth level

Fifth level

44

Now, the main problem:

Can the performance of the Semantic Technologybe ever enough for Big Data?

The current Semantic Technology is often too sloweven for a single-database-size datasets, and nowwe need to handle Big Data!

Should we just give up?

Page 12: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Click to edit Master title style

Click to edit Master text styles

Second level

Third level

Fourth level

Fifth level

55

We believe:

The Semantic Technology can be relevant in the BigData age.

It may be a time, however, to give up on the idea of atriple-store (RDF database) ?

Semantic Data Virtualization is the way

Allowing data to be stored in a non-semantic form.

Providing a semantic view on these data.

This means query transformation instead of datatransformation.

VTT’s DataBearings system follows this approach, shows its feasibility(https://sites.google.com/site/databearings/)

Main open issues: How to enable other semantic functions, such as OWLreasoning, on virtualized data?

Page 13: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Click to edit Master title style

Click to edit Master text styles

Second level

Third level

Fourth level

Fifth level

6

TECHNOLOGY FOR BUSINESS

Page 14: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Handling and Modeling Data onSmart TransportationManagement Systems

Jason Gu and Umar Farooq

Electrical and Computer Engineering

Dalhousie University Halifax, N.S., Canada

Page 15: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

System Components and Data Management

System Components

• Bus Station Module

• In-Bus Module

• Base Station Module

• Bus Stop Module

Data Management

• Artificial Intelligence

• Semantics Technology

• Big Data

Page 16: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Bus Station Module and Data Management

Hardware Modules

• Laser Sensor

• GSM Modem

Data Management

• Less Amount of Data

• Handled Locally

Page 17: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

In-Bus Module and Data Management

Hardware Modules

• GPS Receiver

• GSM Engine

• Microcontrollers

• NV RAM

Data Management

• Less Amount of Data

• Handled Locally

Page 18: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Base Station Module and Data ManagementHardware Modules

• GSM Engine

• Microcontrollers

Data Management

• Large Amount of Data

• Big Data Paradigm

• Semantics Technology

(Query Transformation)

• Nearest Neighbor Algorithm

Page 19: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Bus Stop Module and Data Management

Hardware Modules

• GSM Engine

• Microcontrollers

• NV RAM

• Dot Matrix Display

Data Management

• Less Amount of Data

• Handled Locally

Page 20: #1 sldies intro - IARIA · 2016. 9. 11. · Michel ClauB Technische Universität, Chemnitz, Germany • Challenges for data integration in industrial settings (aspects: standards,

Big Data Analysis and Visualization

• All Computations Performed

at Base Station on Virtualized

Data

• Big Data Routines (Nearest

Neighbor Algorithms)

• Recommendations Generated on

Per Stop Basis