The Semantics of Classification Motivating the New Part 2

1
The Semantics of ClassificationMotivating the
New Part 2

• Jim Carpenter
• Bureau of Labor Statistics

WG2 Meetings Santa Fe, NM January
27-31, 2003
2
Note on this Presentation

• This presentation focuses on the semantics of
  Object X is classified by category Y and
  describes the new attributes to record the
  semantics.
• Other new attributes were discussed in the Open
  Forum tutorial on Part 2. These attributes are
  presented in the UML diagrams at the end of this
  presentation, although they have been updated.

3
Outline

• The nature of classification
• Meaning of the proposition
• Object X is classified by category Y
• Use cases of classification in 11179 MDR
• Attributes of classification schemes that should
  be in a 11179 MDR
• Proposed model of classification region

4
Definition of Classification Scheme

• the descriptive information for an arrangement or
  division of objects into groups based on
  characteristics, which the objects have in
  common.
• Will use for motivation

5
Views of Classificationdescriptive information
for the division of objects into groups

• Process of classifying
• Dividing objects into groups
• MDR should provide descriptive information to
  facilitate this process
• Result of classifying
• Description information that helps us to
  understand the classified item
• MDR should provide descriptive information to
  understand the statement object X is classified
  by the category Y

Following use cases
6
Fundamental Proposition
The MDR should provide the attributes that

• facilitate the classification process
• facilitate understanding the classified item

7
Example of Classification

• Scheme for classifying writing instruments
• Writing instrument an object used for making
  marks on some material
• Differentiating characteristic material in the
  writing instrument that makes the mark
• Material Name of category
• chalk chalk
• ink pen
• lead pencil
• Object X is classified as a pen has meaning

8
Meaning

• Object X is classified as a pen
• is equivalent to
• Object X is a writing instrument that uses ink
  to make marks
• We naturally substitute the category label with
  the description of the category
• Just like a concise terminological definition
• Experiment
• If I give you a funny looking object and tell you
  it is a pen, you know more about it if you know
  about pen.
• If you dont know, then the classification scheme
  provides the descriptive information

9
Object X is classified as a pen

• Object X is a subtype of the class pen.
• But not all classifications are subtype
  relationships

10
Question

• What is the meaning of
• Data element X is classified by the category of
  pen ?
• Answer Data element X is associated in some way
  with objects that are pens.

11
Data element X is classified by the category of
pen

• Data element identifier xxx
• Data element name Pen Ink Color
• Context office supply database
• Object class pen
• Property color of ink
• Data element concept ink color for a pen
• Data element xxx has a data element concept with
  the object class pen

12
(No Transcript)
13
Use Case Setting
Classify data elements according to the day of
the week in which they were created.

• Objects to classify data elements
• Characteristic used to classify the day of the
  week in which an object was created
• Category names Monday, Tuesday, Wednesday,
  Thursday, Friday, Saturday, Sunday
• Category type description
• day of the week in which an event occurred

14
Common Semantics

• Day of the week is a time interval
• A time interval describes an event (occurrence)

15
Use Case Setting, cont.

• Data element - Identifier 111
• Data element - Creation date Monday, January
  27, 2003

16
Classification Scheme 1
Use Case 1

• Classification Scheme Identifier 0001
• Classification Scheme Name Days of the week
  
• Classification Scheme Type name List
• Category Values Mon, Tue, Wed, Thu, Fri,
  Sat, Sun
• Category Type name abbreviation for the name
  of a day of the week (when an event occurred)
• Result the proposition data element 111 is
  classified by the Category with value Sun
• D.E. 111 is associated with an
  event that happened on a Sunday

17
Missing Information
Use Case of current MDR model

• Known by the classifier
• Data element creation date
• Data element 111 was created on a Wednesday.
• was created on is an attribute of the
  classification relationship
• to be supplied by the classifier

18
Conclusion 1

• The classification relationship in ISO 11179
  needs an attribute
• The attribute should describe the characteristic
  of the administered item used by the classifier

19
(No Transcript)
20
Classification Scheme 2
Use Case 2

• Classification Scheme Identifier 0002
• Classification Scheme Name Days of the week
  
• Classification Scheme Type name List
• Category Values 1, 2, 3, 4, 5, 6, 7
• Category Type name Numerical designation for
  a day of the week (when an event occurred)
• Result the proposition data element 111 is
  classified by the Category with value 7, which
  represents a day of the week

21
Missing Information

• Attribute value of the classification
  relationship
• Creation date
• Meaning of the value 7
• Which day of the week is 7?
• Could be provided by a reference document
• Inconvenient
• Not machine readable

22
Conclusion 2

• Need an attribute of Category that states the
  meaning of the Category value

23
Complex Categories

• Complexity from users viewpoint
• Multiple characteristics
• Fuzzy boundaries
• Technical terms
• Numerous types and instances of category
  relationships

24
Conclusion 3

• Users need more than concise terminological
  definitions of categories
• Included concepts
• Excluded concepts
• Examples
• Illustrations (picture, sound, animation, )

25
Non-English Language Speaking User

• All information is missing
• Need translations of attribute values

26
Conclusion 4

• Designations and definitions of terms should be
  represented in the languages of the MDR users

27
Proposed Model Summary of Changes

• Make category an administered item
• Expand Terminological Entry class
• Signs language independent
• More definitional types
• Make classification relation an administered item
• Other useful attributes

28
Note The following diagrams use inheritance as
discussed this morning. Administration_Record is
removed as an attribute from all
Administered_Item types
29
(No Transcript)
30
(No Transcript)
31
(No Transcript)