Module 8.2: Representation – Classification, Categories, Taxonomies, Ontologies

LIS 5043: Organization of Information

Dr. Manika Lamba

Introduction

In this module, we are going to examine another way in which we provide access to information and knowledge resources that is also related, in most instances, to the subject(s) of the objects.

There are various structures for organizing knowledge that have been developed in library and information science, as well as other disciplines, such as natural sciences (biology, zoology), neuroscience, cognitive science, etc., as well as in professional and commercial contexts.

The most common of these structures for organizing knowledge is classification schemes (ie. Dewey Decimal Classification scheme, Library of Congress Classification scheme). Other related concepts that we will explore in this module are taxonomies, and ontologies, which are very closely related to classification schemes, and are often structured in a similar manner.

In this module, you will explore classification as a broad concept, but also use as an example, the bibliographic classification schemes of DDC and LCC in order to understand the purposes and uses of classification for subject access.

You will extend your thinking to taxonomies, ontologies, and also folksonomies, by exploring how classification is used on the web, in e-commerce, and in other non-LIS contexts, such as in the development of corporate taxonomies and ontologies.

Classification: Philosophical Roots

• “Classical Classification” is based on the Aristotelian idea of categories and class membership
• Is a binary relationship: you either belong to the class or you do not belong to the class
• Membership is based on similarity (having the same characteristics)

Classification or the idea of “classing” or grouping concepts or objects (people, places, things) together based on similarity is based on Aristotle’s “classical theory of classification“. The classical view of categories is that concepts or things are grouped together into a category based on some similarities or common properties. As we move through our lives we automatically categorize people, animals, and things into categories.

According to George Lakoff, there is nothing more basic to our thought, perception, action and speech than categorization. Whenever we reason about a kind of thing, such as a chair or an emotion, or whenever we intentionally perform any kind of action, for example sorting laundry, we are employing categories that we have previously developed based on past experiences and motor actions. Categorization is “central to any understanding of how we think and how we function, and therefore central to an understanding of what makes us human” (Lakoff, 1987: 6).

Classical classification is a binary relationship: you either belong to the category, or you do not. So membership is based on common attributes and if you do not possess these common attributes, then you do not belong to the category.

Aristotle’s “classical theory of classification” influenced library and information science classification schemes and is evident in the definitions from library and information science literature.

Classification, as defined by Taylor and Joudrey, is “a structured system of categories used to collocate similar ideas or objects.” (2009: 383) For Elaine Svenonius, classification “brings like things together….like things are brought together with respect to one or more specified attributes” (Svenonius, 2000: 10).

Beginning in the sixteenth century, and also parallel to the growth of libraries, many different bibliographic classification schemes (those used for retrieval in bibliographic systems) were developed based on the classical theory of classification.

Those of significance were based on either the philosopher Francis Bacon’s division of knowledge into three basic “faculties”: history (natural, civil, literary, ecclesiastical); philosophy (including theology); and works of imagination (poetry, fables, etc.), or on the classification work of Thomas Jefferson.

Other bibliographic classification schemes based on Aristotle’s classical theory include those still in use today, the Dewey Decimal Classification (DDC) scheme and the Library of Congress Classification (LCC) scheme, as well as others mentioned earlier.

Classification: Philosophical Roots

• Prototype Theory (Wittgenstein and Rosch)
  • Wittgenstein disputed the classical category theory around 1953 (gameness)
    • Family resemblances
  • Rosch and other cognitive psychologists further developed what is now called the “Prototype theory” of categories/categorization
    • Categories based on human experience, imagination

Many systems for organizing knowledge are based on the idea of categories or classing items together in either branched or hierarchical structures. The categories in structures for organizing knowledge, for example thesauri or classification schema, are based on understandings of the world’s knowledge and how each class is related to the other classes contained in the whole structure. Categories and the formation of categories are central foci of what cognitive scientists study.

Cognitive science is a relatively new field that brings together many academic disciplines that focus on developing a greater understanding of the human mind and how it works. The disciplines include psychology, linguistics, anthropology, philosophy, and computer science.

According to George Lakoff, cognitive science seeks detailed answers to questions such as: What is reason? How do humans make sense of their experiences? What is a conceptual system and how is it organized? Do all people use the same conceptual system? Are there commonalities in the way that humans think? (Lakoff, 1987: xi).

The classical view of categories or classical classification remained the predominant view in organizing knowledge until 1953 when Ludwig Wittgenstein pointed out that some categories did not fit the classical mold. He used the category of game as an example of how the boundaries (the properties the things in the category must have in common in order to belong to the category) that define gameness are not the same for all games.

Some games are competitive, such as football or board games, and other games require skill or strategy, for example chess, yet still others are purely for amusement, such as ring-around-the rosy. Wittgenstein proposed, instead, the idea of “family resemblances”, in which members of a category resemble each other in various ways, but may not share a common set of properties.

While Wittgenstein’s “family resemblances” was one of the first challenges to the classical view of categories, Eleanor Rosch, has been credited with developing the theory of prototypes and basic-level categories or “prototype theory”. Prototype theory proposes that human categorization is based on both human experience (of perception, motor activity, and culture) and imagination (of metaphor, metonymy, and mental imagery).

Prototype theory challenges the classical view of reason as disembodied symbol-manipulation and the mind-as-computer (mind as symbol manipulating machine) metaphors. The prototype theory also calls into question the idea of “What is a category?”

As noted earlier we develop categories for things, and it is through these categories that we make sense of the world around us. What happens to these understandings of the world if we change our definition for what constitutes a category? (Lakoff, 1987).

These questions continue to be explored by researchers who support prototype theory and later, the concept of cognitive models.

Themes of Prototype Theory

Central themes related to prototype theory that might be important or related to structures for organizing knowledge includes:

1. family resemblances, or the idea that members of a category can be related to each other without possessing all of the same properties that define the category
2. centrality, or the idea that some members of a category may be ‘better examples’ or exemplars of the category than others
3. gradience, or that some categories have degrees of membership or no clear boundaries for membership
4. basic-level categorization, the idea that categories are not merely organized in a hierarchy from the most general to the most specific, but are also organized so that the categories are cognitively ‘basic’ are in the middle

Some of the central themes of prototype theory are listed on this slide.

As you begin to look at different ideas related to classification and at examples of classification schemes, keep in mind that most classification schemes in predominant use today are based on the “classical theory” of classification, not on the prototype theory.

Also, envision:

• How schemes built around prototype theory might work in information retrieval systems?
• Would they be more effective than those based on classical classification?
• What would some of the benefits be?
• What issues would be present for developers of classification systems?
• How do folksonomies and ontologies reflect the prototype theory of classification?

Definitions

Category(ies)

a group of objects, concepts, etc. that are related in some manner; by similarity, sharing some attribute or characteristic

Classification

arrangement of entities or concepts in logical order according to their similarities

Taxonomy(ies)

sets of categories with hierarchical and other semantic relationships between categories shown; support classification

Ontology(ies)

a formal, explicit representation of a domain’s knowledge; a map of the knowledge of a domain or specialization, containing terminologies as well as semantic relationships

Now that we have learned the “roots” of classical classification or the basis for many classification schemes, let’s look at a few definitions you will encounter in your readings for this module.

Taxonomies, ontologies, classification are three of the most difficult terms to differentiate when we discuss the organization of information and knowledge resources. Be sure that you read Taylor and Joudrey’s definitions and examples of how each are related and used.

One important distinction to make between taxonomy, ontologies, and bibliographic classification is that bibliographic classification is concerned with classifying and representing the entire world of knowledge.

Taxonomies and ontologies, generally are representing a specific domain or context. For example, a corporate ontology, is usually developed to be used in a specific corporation. It contains terminology used by the people of the corporation, their specific field, and is related to a specific set of problems, products, etc.

It’s structure is based on a set of relationships, nodes, etc. that make sense in that specific context. It also may be so highly customized that it is not useful beyond the context in which it was developed, but can be extended to other problems, products, etc. of similar nature.

In this module you will be taking a closer look at the concepts of classification, taxonomies, and ontologies.

Definitions

Classifying

1. Arranging collection according to classification system
2. Assigning notation or class number to each item

Class

group of entities or concepts possessing common attributes or characteristics

Facet

component of a complex subject based on a particular characteristic, e.g., geographic facet, language facet, literary form facet

Scheme

formal system for arrangement of entities according to subject or form

Some more definitions to consider.

A facet is also a related concept that you will be introduced to in this module. A facet is “a component (or part of) a complex subject based on a specific characteristic”. In LCSH you reviewed subdivisions related to topic, geographic location, chronological aspect, or form. These can be viewed as facets. We will look at what is called “faceted classification” later in this Introduction.

Scheme is also an important term to learn. It is the formal system of recording the classification structure.

Definitions

Schedule

published description of classification scheme showing overall structure and relationships among subjects or concepts

Index

• alphabetical list of subject terms and concepts with locations in scheme
• cross-references for subjects scattered by scheme

Notation

• symbols representing classes and subclasses in scheme
• applied as a mark on an item identifying item with its class

Notation characters

Pure: Only numerals or only letters

Mixed: Letters plus numerals; alphanumeric

More definitions to keep in mind.

Classification’s end product, is a notation that shows how the object is classified, or which class it fits into. There are several forms of notation, as described here.

For example, the DDC uses a pure numerical system of notation. LCC uses mixed alphanumeric system.

Unique Identification

Document classification codes are control devices that serve several purposes…

• Uniquely identify physical items in collection
• Link physical items with records describing them
• Facilitate intellectual access
• Facilitate physical access
• Facilitate inventory control

In bibliographic classification, one of the goals/purposes is to uniquely identify the object within the collection so that the user can locate/access the object.

The notation helps to provide a mechanism not just for unique identification, but also for physical location of the object. Classing an object also provides for intellectual access, based on similarity, usually similar subject matter.

Unique Identification

Examples

• Notation: Describes item; reflects classification

• Call number: Notation + unique book number

• Accession number: Unique identification by order acquired; arbitrary

Libraries use several forms of notations such as classification codes, but also accession numbers which are written on the physical object but also noted in the library bibliographic record.

Barcodes that you see on library materials also provide this same function.

• What about UPC codes on products in stores?
• Do they uniquely identify items in the store’s inventory?
• Does a UPC serve the same function as a classification notation?

Major Approaches

Enumerative

subjects and their relationships are prearranged in classes and subclasses

Faceted

Potential facets for subject classes are predetermined, but classes and subclasses are not prearranged

In bibliographic classification, there are two major approaches, the enumerative approach where all subjects of the entire world of knowledge and their semantic relationships (BT, NT, RT, etc.) are prearranged in classes and subclasses and the faceted approach, where potential facets are predetermined, but classes and subclasses are not prearranged.

Enumerative schemes tend to be hierarchical in nature so often they are referred as hierarchical schemes. The two most predominant schemes in use worldwide are the Dewey Decimal Classification (DDC) scheme or its derivative Universal Decimal Classification (UDC) and the Library of Congress Classification (LCC) scheme. These two schemes are enumerative, hierarchical schemes.

We will look at some examples of faceted schemes in the module, but the most well known of these is Ranganathan’s Colon Classification scheme, which is gaining popularity in the web community.

Enumerative Approaches

• Subjects and their relationships are prearranged in hierarchical classes and subclasses

• Classification consists of identifying location of each item in scheme.

• Notation for each class is predetermined.

These next few slides summarize both major approaches and give some examples of each. Read the slides for a brief introduction to each. This module will help you explore each in more depth.

Dewey Decimal Classification (DDC)

DDC is an example of an enumerative scheme. It was first published by Melvil Dewey in 1876 and continues to be developed and edited. It is used in many small and medium sized libraries.

Dewey Decimal Classification (DDC)

There are advantages and disadvantages to every system of classification. DDC is often seen as too restrictive and arbitrary, and also too Anglocentric.

DDC: Ten Main Classes

Here is an example of the DDC main classes. DDC in effect, carves up the world of knowledge into 10 classes or groups of subjects related in some manner, or under some common heading.

Where do you think Library and Information Science falls? Knowledge Management?

DDC: Ten Divisions within 000

The ten classes are further subdivided into ten divisions. LIS is classed in the 020 division of the main class of Computer science, information, and general works.

Does this seem the logical place for LIS?

DDC: Ten Sections within 000 Division

Each division is further subdivided into ten sections. LIS breaks down as shown above.

DDC: Subsections within 020

The sections are then further subdivided into subsections. Again, LIS subsections are shown on the slide.

There are of course, additional ways that notations can be developed and classes determined in DDC. These slides do not show the somewhat faceted approaches that are possible using the DDC. The readings explain these options more in depth.

Library of Congress Classification (LCC)

The second predominant enumerative scheme is the LCC. It was developed and published by Herbert Putnam in 1898 and continues to be revised and developed. It is primarily used in medium to large academic libraries, though some public libraries use it also.

Library of Congress Classification (LCC)

It uses a mixed notation. The main disadvantage of this scheme is that it is quite large and can be a bit complex when trying to classify objects with complex subjects.

LCC Example: AC Class

Here is an example of how LCC is structured. This slide shows the AC-AP basic class structure.

LCC Example: Class AC

The AC class is also further subdivided.

LCC Example: Class AC

AC further subdivided. Again, these slides illustrate the basic structure of the LCC. Refer to the readings to learn more about how LCC is structured and applied.

Faceted Approaches

Faceted Classification

• Potential facets for subject classes are predetermined, but classes and subclasses are not prearranged

• Classification consists of identifying each facet applicable to an item

• Notation is synthesized by drawing together notation from different facets

The second major approach of classification is the Faceted Approach.

Using this approach, it is the facets or attributes of the objects that are first identified, but classes and subclasses are not prearranged. When classifying, the facets contained in the object are first identified, and then matched to the facets or foci of the scheme.

Faceted Approaches

Here is a basic example of a faceted scheme for a literature collection. The faceted approach can be used for any collection. Often you will see it used in e-commerce.

Let’s look at how it works.

Each column represents a potential facet of the objects in this collection. Under each facet are the subfacets or what we call foci. Under the Language facet, foci include: American, English, etc.

The facets are not prearranged and the facets and foci can easily be expanded to add on new facets/foci as needed.

Faceted Approaches: Exercise

Here are few examples of how a faceted scheme notation is built. Flip back to the previous slide to see how the scheme was applied to each example.

Colon Classification

Colon Classification in an example of faceted classification.

This system of classification, and really the idea of faceting, was developed by Ranganathan and published in 1933. It has never really been used in a library setting but remains the basis for the development of all faceted schemes in use currently. The web community “discovered” Ranganathan about 13 years ago and have been developing faceted schemes within many e-commerce sites, and within digital libraries.

Colon Classification

PMEST

• Personality (primary characteristics, “essence”) -Matter (physical characteristics)
• Energy (operations, problems, processes)
• Space (geographical, topological)
• Time (date, period)

Index - Included in single volume

Notation - Mixed;Noted for use of punctuation (colon); Includes Roman and Greek characters

The facets of Colon Classification include:

PMEST, as described on the slide.

It includes a very complicated notation with a mixture of characters including colons, dashes, slashes, etc and both Roman and Greek characters.

Colon Classification

The primary disadvantage of Colon Classification is the long notations that result from its application.

Art and Architecture Thesaurus

• AAT provides controlled vocabulary for the visual arts and functions as a classification scheme

• History – Created by many people through Getty program

• Current use – by archivists, slide/photo librarians, museum curators, indexing services, architecture/design firms, art dictionaries and encyclopedias

For:

• Physical description of museum objects, slides, photos, archival materials, etc.

• Subject cataloging and keyword indexing of books, images, periodical literature

• Database searching by scholars, researchers, students, practitioners, librarians

• Schedule: available online

• Index: Thesaurus in an index

• Notation: Mixed;Each descriptor (controlled term) has unique notation code

The AAT, which is a controlled vocabulary you learned about in the 7.1 module, can also function as a classification scheme. The next few slides discuss it use as a faceted scheme.

If you have not already reviewed this controlled vocabulary, here is a link to it . Take a look! It is a very good example of a faceted scheme that is currently in use.

AAT

There are of course advantages and disadvantages to its use. One disadvantage is that it is not designed for shelving or physical location of objects within a collection. In the electronic environment, however, this is not a disadvantage and this scheme is actually very useful in helping users precisely locate objects in information retrieval systems.

As you complete the activities for this module, and also just surf the web, look for examples of faceted classification schemes.

Limitations to Classification

1. Knowledge never remains static
2. Physical placement distorts cross-relationships
3. Systematic organization produces artificial symmetries
4. Non-logical ordering is sometimes preferable
5. Classifiers have psychological tendency to skew system
6. No collection ever precisely reflects universe of knowledge

As with any structure for organizing knowledge, there are limitations to consider, as noted on this slide.

Because knowledge never remains static, classification schemes, taxonomies, and ontologies, also have to be continually reviewed and revised to remain relevant and useful.

In schemes that are used for physical placement, there can be issues related to classing complex subjects, or where to class an object that is about more than one subject? Or does classing an object in a specific class produce an artificial symmetry that may be inaccurate? Or could it in turn lead to a new way of thinking?

Classifiers also have a psychological tendency to skew the system, as they will class items based on their own understandings of the world, a discipline, etc.

Careers related to Classification, Taxonomy, Ontology

• Cataloging and classification
• Ontologists
• Taxonomists
• Information Architect
• Semantic Web

Careers related to the development and application of classification schemes in libraries of course include becoming a cataloger, but have you thought about a career as a classificationist or someone who develops or edits existing classification schemes?

Or as an ontologist or taxonomist who develops specialized ontologies or taxonomies within corporations or for e-commerce, such as an information architect? How do knowledge managers apply the theories of classification/categories within their specific contexts?

Take a look also at the Semantic Web and Knowledge Organizing Structures (KOS) being developed there.