Standards for Scientific and Technical Data

John Rumble

National Institute of Standards and Technology

President, CODATA

john.rumble@nist.gov

Standards for Scientific and Technical Data

•     Needs for S&T data and database standards
•      Basic approaches to standards
•      Content of S&T data and database standards
•      Types of standards bodies
•      Examples

•    Where does CODATA fit in

Needs for S&T Data and Database Standards

• Improved data collection

      completeness, uniformity, easier  

• More efficient database building

      collective solutions to nomenclature and representational problems  
      basis for data dictionaries  

• Data exchange and integration

      sharing, combining, and comparing data
• Easier data use

      common connections to applications, knowledge discovery

Basic Approaches to Data and Database Standards

• Neutral formats
  
      allows one to maintain own format
      only requires translation into and out of format
• Define data elements
        include normal data dictionary information  
        identify minimum set necessary
       define others in case wanted or needed
       allow extensions (self-definition)
       Do not underestimate difficulties in resolving nomenclature problems
  
• Separate semantics from syntax

Basic Approaches to Data and Database Standards

• Neutral formats
       allows one to maintain own format
       only requires translation into and out of format
  
• Define data elements

• Separate semantics from syntax

Content of S&T Data and Database Standards - 1

• Description of substance, systems, species, taxa,
  
• Reporting properties, measurements, observations, characteristics, calculations, results
  
• Property context

Description of Substances, Systems, Species - 2

• Required for computerized description
  
• Uniqueness
        it is this substance, not that one
        different levels - protein, protease, protease IV, protease IV (acidopolus), ...
  
• Equivalency
        same to specified levels all ketones,
        all ethyl-ketones, R- ketones, …
  
• Can include association (bonding, joining, etc.), interactions, reactions

Description of Substances, Systems, Species

• Several types of information

  primary identifiers (names), specifications, characterization and composition, source, processing history, reference test results, association(bonding, joining), material form, supplement information

• Millions of chemicals, species, people,
  
• Do not have to include all in standard
  
• Often have variety of description approaches
  
• Most S&T substances, systems and species need information modeling -I.e., for biological species

Reporting Properties, etc.

• Properties, measurements, observations, characteristics, calculations, results
  
• Usually more complex structure than realized
  
• Greater dependency on variables than thought
  
• Often can be represented in a variety of ways - text, numbers, equations, coordinate systems, other
  
• Multiple nomenclature systems also a problem here

Property Context -1

• Conditions (situation) under which the property, observation, etc. is meaningful
  
• One meaning of metadata
  
• Independent variables of two types
        those set at beginning and not changed
        those varied throughout data set
  
• Includes data collection methods, data analysis documentation, others

Property Context -2

• Variables can be very numerous (hundreds) and complex
• Information models help greatly
• Most researchers do not record every variable
        Must construct definitions of two types
        mandatory - without these data set would be useless
        optional - if you record, do it this way
• List
        controlled vocabular
        suggested vocabulary
        free text

Types of Standards Bodies

• Formal
        ISO
        EU
         National Standards Development Organizations (SDOs, e.g., JSI))
• Informal
        Professional and technical societies
        Unions
        Pre-standardization
  
• New
        OMG
        W3C
         CODATA

Examples of S&T Data and Standards

• Crystallography - CIF, mmCIF
  
• Physics - neutron collisions, nuclear structure
  
• Analytical chemistry - IR, NMR, Mass Spectra
  
• Chemical nomenclature - IUPAC, CAS, CML
  
• Surface characterization - XPS
  
• Engineering materials - ISO STEP, ASTM, matML Biology - viruses, … Molecular
• biology - mmCIF, OMG

The Standards Process

1. Someone identifies need for standard
2. A group is convinced of the need
3. A technology (approach) is proposed
4. A consensus is achieved by all interested parties
5. The standard is published
6. The standard is used by people and groups who have a vested interest (business reason) to do so

Some Standards Economics and Sociology

• You must be motivated to build a standard
  
• You must be motivation to use a standard
  
• Industry builds and uses standards for business reasons
  
• If it is hard to build a standard, the motivation is usually lacking

Standards and Science

• Scientists are often reluctant to use standards because they are not the state of the art
  
• Scientists are not accustomed to working towards a consensus
  
• Scientists are usually uncomfortable in working with a formal standards group
  
• Most scientific data standards are developed because of a hidden business motivation

Where does CODATA fit in - 1

• Long experience in development of data reporting requirements
• Multi-national, multi-disciplinary
• Existing Commission, Task Group activities
• Friendlier environment for scientists
• Considerable experience in setting scientific data standards

Where does CODATA fit in - 2

• Instructional and training materials
• Task groups
• Web and print publication
• Neutrality
• Cross-union cooperation
• Comprehensive survey of scientific data and database standards
• CODATA wants to create an environment helpful for data and database standards