Indexing with a Controlled Vocabulary

1
Indexing with a Controlled Vocabulary

• Basic Concepts

2
Indexing Topics Covered

• The concept triangle
• The five-axiom theory of indexing
• The indexing process

3
The Concept Triangle
Referent
Expression
Concept
4
The Referent

• The referent is everything about which a
  meaningful statement can be made.
• For example, about a certain table many
  statements can be made concerning the material of
  which it is made, its price, purpose, producer,
  weight, the structure of its surface, etc.

5
The Concept

• We define the concept as the sum of the
  essential statements that can be made about a
  referent.
• Essential statements are those which contribute
  to the characterization of the referent itself.
• Inessential statements are those which do not
  contribute to the characterization of the
  referent itself.

6
Kinds of Concepts

• General concepts
• The general concept describes a class of
  interrelated referents.
• For example metal, oxidation, information
• Individual concepts
• The individual concept is one to which no
  meaningful conceptual feature can be added.
• For example Albert Einstein Fritz the Cat.

7
General vs. Individual Concepts in Indexing

• It is the task of subject indexes to provide
  access to documents or text passages relevant to
  general concepts.
• An information system which works quite well for
  individual concepts, may totally fail when it is
  required to manage general concepts too.

8
The Mode of Expression

• Lexical expressions
• linear strings of characters commonly agreed upon
  to express concepts or concept connections
• Non-lexical expressions
• linear strings of characters by which concepts or
  concept relations are expressed and upon which no
  firm agreement has been made

9
Forms of Expression Indexing

• Lexical expressions require little indexing work
• Often appear in Identifier fields rather than in
  Descriptor fields of database records
• Non-lexical expressions require indexing work
• non-lexical expressions exhibit ambiguity and
  multiplicity

10
Concepts Expressions

• Individual concepts are almost always expressed
  lexically
• General concepts are almost always expressed
  non-lexically
• In natural, uncontrolled language there is an
  unlimited multitude of non-lexical, paraphrasing
  expressions for concepts
• Multiplicity ambiguity of natural language
  expressions are largely restricted to general
  concepts

11
Five-Axiom Theory of Indexing

• Definability
• Order
• Sufficient degree of order
• Representational predictability
• Representational fidelity

12
Axiom of Definability

• The compilation of information relevant to a
  topic can be delegated (to a skilled specialist
  or a programmed search mechanism) only to the
  extent to which the inquirer can define the topic
  in terms of concepts and concept relations.

13
Axiom of Order

• Any compilation of information relevant to a
  topic is an order-creating process.
• Order is defined as the meaningful proximity of
  the parts of a whole at a foreseeable place.

14
Axiom of Sufficient Degree of Order

• The demands made on the degree of order increase
  as the size of the collection and/or the
  frequency of the searches and/or the specificity
  of the searches increases.

15
Axiom of Representational Predictability

• The completeness of any search for documents
  relevant to a topic of interest depends on the
  predictability of the modes of expression for
  concepts in the search file.
• Successful searches require a language with
  predictable modes of expression for concepts.

16
Axiom of Representational Fidelity

• The precision of any search for documents
  relevant to a topic of interest depends on the
  fidelity with which the modes of expression for
  concepts can be expressed in the systems
  language.

17
The Indexing Process

• Step 1
• Determine the essence of a document
• Step 2
• Represent this essence with sufficient degrees of
  predictability and fidelity

18
Importance of Categories

• The predictability of essence selection is
  markedly enhanced when the indexers have an
  orientation to conceptual categories.
• For example, in some chemistry databases, all
  descriptors belong to the following categories
• MATTER
• LIVING ENTITY
• APPARATUS
• PROCESSS
• In ERIC, the nine Descriptor Groups serve as
  categories.

19
Natural Language Indexing

• Natural language expressions, as derived from
  original texts, can only in the case of
  individual concepts lead to an information system
  of adequate quality and survival power.
• The specificity of natural language expressions
  is compromised by their lack of predictability.

20
Importance of Cutters Rule

• Precise and complete searches require that the
  most specific descriptors that the vocabulary
  provides be chosen for the indexing of a subject.
• A query with a specific descriptor must not
  retrieve concepts that are more general than the
  search descriptor.

21
Importance of Syntax

• In the interests of enhanced representational
  fidelity any advanced indexing language needs a
  syntax in addition to its vocabulary.
• The syntax should represent the manner in which
  the concepts are connected with each other in the
  texts to be stored.