Industrial data quality

1
Industrial data quality
Guy Pierra pierra_at_ensma.fr
2
Content

• Industrial data quality dimensions
• Industrial data quality factors
• IDQ effort within WG2
• Conclusion / Recommendation

3
Industrial Data Quality dimensions
Data Quality dimensions? 1 - Data correctness
(intrinsic) It is generally agreed that data
should also "fit for their intended used" 2 -
Accessibility 3 - Relevance 4 - easy to use
Abstraction process
symbols
Business situation
Quality factors How to improve quality?
4
Industrial Data Quality factors
Data Quality dimensions? 1 - Data correctness
(intrinsic) It is generally agreed that data
should also "fit for their intended used" 2 -
Accessibility 3 - Relevance 4 - easy to use

• 1 - Check Data Correctness (CDC)
• Method? Compare with other data

Abstraction process
? ? ?
symbols
Business situation
?

• Cost/benefit analysis
• Application domain ? ?
• Cost
• specifying ?
• checking ?

Quality factors How to improve quality?
Cost ? Benefit?
5
Industrial Data Quality factors (cont.)

• 2 - Improve abstraction Process (IP)
• Method? ISO 9000-like certification

Data Quality dimensions? 1 - Data correctness
(intrinsic) 2 - Accessibility 3 - Relevance
4 - easy to use

• Application domain
• Data correctness ?
• Accessibility ?
• Relevance ?/?
• easy to use ?
• Cost
• specifying ? ?
• checking ? ? ?

Abstraction process
?
? ? ?
Quality factors How to improve quality?
symbols
Business situation
Cost ? ? ? Benefit ?
6
Industrial Data Quality factors (cont.)

• 3 - Improve intrinsic data quality (IDQ)
• Method? Concentrate on model issues
• Data correctness Improve data semantics

Abstraction process
?
symbols
Business situation
7
Industrial Data Quality factors (cont.)

• 3 - Improve intrinsic data quality (IDQ)
• Method? Concentrate on model issues
• Data correctness Improve data semantics

• Set-theoretic view of data
• an entity, an attribute stands for a set of
  values
• restricting the size of the set through formal
  constraints avoid errors

8
Industrial Data Quality factors (cont.)

• 3 - Improve intrinsic data quality (IDQ)
• 3.1 - Data correctness Improve data semantics
• SC4 application
• maintain/extend use of constraint-based
  modeling
• Develop constraint-based usage guide (e. g.,
  SASIG) ? ?
• 3.2 - Accessibility
• SC4 application
• Ensure free accessibility when necessary
  condition of usage ? ?
• 3.3 - Relevance
• SC4 application
• Consider developing formal usage guide / user
  requirements ?
• 3.4 - Ease of use
• How to? Major ease of use problem lack of
  interoperability!!
• SC4 application
• Ensure interoperability of SC4 standard !! ?

9
Industrial Data Quality factors (end)
1 - Check Data Correctness (CDC) 2 - Improve
abstraction Process (IP) 3 - Improve
intrinsic data quality (IDQ)
Cost ? Benefit?
Cost ? ? ? Benefit ?
Cost ? Benefit ? ?
10
IDQ Quality effort within WG2

• 3.1 - Data correctness
• The PLIB dictionary/ontology meta-schema includes
  a huge number of constraints
• Part 42.2 contains a constraint schema for
  instance values
• 3.2 - Accessibility
• A lot of actions to ensure free-of-charge access
  to PLIB dict.
• Development of a set of Web services for Internet
  access
• 3.3 - Relevance
• nothing!!
• 3.4 - Ease of use
• A world-wide initiative to ensure
  interoperability of all product domain
  dictionaries OIDDI
• Collaboration with UN/CEFACT to ensure
  interoperability/ orthogonality with Business
  Process

11
Conclusion / Recommandation

• Data Quality spreads over a number of dimensions
• From the three identified factors IDC, ID, IDQ,
  Intrinsic Data Quality seems both
• the most efficient from a cost/benefit analysis
• the most adapted to SC4 technology
• Main recommendations to SC4
• promote the development of formal
  constraint-based usage guides
• request specific free of charge accessibility for
  those SC4 standards that defines
  computer-sensible meaning identifiers
• ensure interoperability between SC4-defined
  industrial data standards