Dr. Tom's Classification Guide

Thomas D. Wason, Ph.D. (aka Dr. Tom) http://www.tomwason.com [Home] wason@mindspring.com One of the Dr. Tom Guides

Purpose of Document

The purpose of this document is to provide guidelines for the use of the classification category in the IMS Meta-data and the IEEE LOM (Learning Object Metadata). This document includes an explanation of the rationale behind classification.

Document Information

Contents

1. Introduction
2. Classification Purpose
3. Structure of the Classification Category
4. What is a taxonpath?
5. Definitions of the Purpose Vocabulary Terms
6. Why aren't more fields executed using Classification?
7. An example: Discipline (Subject)
8. An example: Educational Difficulty
9. Conclusion

classification provides elaboration and extension of the meta-data.

Elaboration:
The meta-data without classification was considered to provide inadequate vocabularies for many of the fields. For example, the field of difficulty under the category educational provides only five levels (0..4) of difficulty. What do these five levels mean? Under what context is the difficulty to be interpreted? A learning module on algebra that may be difficult for a 12-year old may be of only intermediate difficulty for a 15-year old. Clearly, educational/difficulty does not provide the richness that many desire. There is no widely adopted system for defining difficulty, however. The use of classification enables elaboration of the difficulty field.

Extension:
A module may be designed to serve specified educational objectives. Not only is there no well-defined place to put descriptions of the educational objectives, there is also not a globally adopted descriptive system for educational objectives. The use of classification enables extension of the meta-data to include description of a resource in terms of the educational objectives supported.

There are two critical terms in understanding the classification category: purpose and description. The overall purpose of the classification category is to provide descriptions of the resource. A purpose of the classification is to provide a place to put information that describes the resource in a manner not provided in the meta-data structure elsewhere. The descriptions are captured within taxonpaths, the description and the keywords.

The classification purpose always refers to the purpose of the classification, not the purpose of the resource.

The following purposes have been defined:

• Discipline
• Idea
• Prerequisite
• Educational Objective
• Accessibility Restrictions
• Educational Level
• Skill Level
• Security Level

These purpose terms are defined below. Another purpose of the classification category is to provide a place to refine descriptions provided elsewhere in the meta-data. These are purposes that have not been previously defined.

classification provides descriptions of the resource. Classification does NOT use the purpose to define purposes of the resource. It may be convenient to mentally substitute the word "type" for "purpose". "purpose" was not used in the specification because "type" is already used heavily.

The core issue is the differentiation of educational meta-data different from general-purpose meta-data. It is in the core educational areas (and include rights management here, too) that we can expect to see growth and change through practice. The field of managed online education is in its infancy. Thus, it was felt necessary to address the need in the two-level approach: simple scales and vocabularies, and flexible classifications. The fact that these core descriptions seem to have such sparse representation in the meta-data does not in anyway indicate that they were felt to be unimportant, or that they were not considered: quite the contrary. Instead, a consistent strategy for handling these recurrent problems was developed.

The classification category is a structure. Each classification has a purpose: what sort of description resource is being provided? It may contain descriptive terms from an organized system of description in one or more taxonpaths. It may provide a brief textual description and/or a set of keywords relative to the specific purpose of this particular classification. The structure of the classification category is as follows:

This can also be represented in outline form:

classification
	purpose
		langstring: [value]
	taxonpath
		source: [value]
		taxon
			id: [value]
			entry
				langstring: [value]
	  [taxon*
			id: [value]
			entry
				langstring: [value]]
	description
		langstring: [value]
keywords
	langstring: [value]

In order to explain a taxonpath, let's look at the structure of most taxonomies. A taxonomy is an organized structure of information or terminology. Most taxonomies are constructed as hierarchies. At the top level is a list of broad terms. Under each term is a list of terms that refine each of the broad terms. And then under each of those terms might be another list. This list of lists of lists can have significant depth. Some taxonomies have up to nine (9!) levels. For more information on taxonomies, see "Dr. Tom’s Taxonomy Guide", drtomtaxonomiesguide.html. A hierarchical taxonomy as an outline would have the structure of:

1
	1.1
		1.1.1
		1.1.2
		1.1.3
	1.2
		1.2.1
		1.2.2
		1.2.3
	1.3
	  . . .
2
	2.1
		2.1.1
		2.1.2
	2.2
		2.2.1
		2.2.2
	2.3
		2.3.1
	 . . .
3
	3.1
		3.1.1
	3.2
	3.3
		3.3.1
		3.3.2
		3.3.3
	 . . .
. . .

2
	2.1
		2.1.1
		2.1.2
	2.2
		2.2.1
		2.2.2

2
	2.2
		2.2.1
		

Taxonomies do not bloom from no place. A taxonomy is created. The creator of the taxonomy is the source. The source is the entity (person, organization) that is responsible for managing the taxonomy.

A specific example from the Library of Congress Classification is:

B -- PHILOSOPHY. PSYCHOLOGY. RELIGION
	F, Psychology
		180, Experimental psychology

taxonpath
	source: LCC
	taxon
		id: B
		entry: PHILOSOPHY. PSYCHOLOGY. RELIGION
	taxon
		id: F
		entry: Psychology
	taxon
		id: 180
		entry: Experimental psychology

The concatenated id is: BF 180.

A taxonpath derived from the library of Congress Subject Heading (LCSH) is:

	<taxonpath>
		<source>LCSH
		</source>
		<taxon>
			<id>B
			</id>
			<entry>
				<langstring lang="en">
					Philosophy, Psychology, Religion
				</langstring>
			</entry>
		</taxon>
		<taxon>
			<id>F
			</id>
			<entry>
				<langstring lang="en">Psychology</langstring>
			</entry>
		</taxon>
		<taxon>
			<id>180
			</id>
			<entry>
				<langstring lang="en">
					Experimental Psychology
				</langstring>
			</entry>
		</taxon>
	</taxonpath>

The classification purpose is the purpose of each particular classification, NOT the purpose of the resource. The list of terms for purpose provided is not exhaustive, but suggestive. First, any field elsewhere in the meta-data may be elaborated by using the field name as the purpose. It is important to remember that classification may be repeated with a different purpose each time, and that multiple taxonpaths may be included within a single classification block.

Discipline: In the US, we would call this "Subject". The term "Discipline" was chosen as being more internationally meaningful. This is directly equivalent to the Dublin Core (DC) Subject: "The topic of the content of the resource" (http://purl.org/DC/documents/rec-dces-19990702.htm). We agree with the DC suggestion that whenever possible, subject terms should be selected from a standard classification system, which is typically a taxonomy. Thus, use of the taxonpath is urged, with the source defined.

Idea: In the US, we could call this "Concept", but in Europe Concept is considered a subdivision of Subject or Discipline. An example of an Idea is Evolution. As a Discipline, this refers to the biological development of species, but as an Idea, it may be applied to economics, sociology, ideas and so forth. Idea, or concept, navigation can be very powerful. Again, the use of standard taxonomies is encouraged.

Prerequisite: An educational term meaning that which is required in order to use this resource. The use of standard classifications and taxonomies is encouraged. Note that the use of the same educational objective classification may be used for both prerequisites and educational objectives. There are those who feel that the coordinated use of prerequisites and educational objects around a common standard may be a good way to automate the selection of resources for an educational experience.

Educational Objective: The objective(s) to be achieved through the use of the resource. Use of one or more standard taxonomies is urged.

Accessibility Restrictions: Perceptual or motor requisites to use a resource. for example, must have normal vision. I would prefer the use of the purpose "Accessibility Enhancements", as suggested by one of your colleagues at Deakin. Thus, a resource that can be used by someone without normal vision because it includes an audio output would have an accessibility enhancement.

Educational Level: The grade level for which the resource is intended. This addresses a can of worms. First, there is no universally accepted grade level system. Each country has its own variant. The wise classifier would use multiple taxonpaths from multiple sources here so that the resource may be discovered and used by the widest applicable audience. This technique of the use of multiple taxonpaths from multiple sources should allow a library to create a thesaurus among taxonomies over time, based on the statistics of application. [This applies to most instances of Classifier.] An additional confound is that many resources will have use at several educational levels, adjusting for difficulty and user populations. The construction of a taxonomy to express these conditions is urged.

Skill Level: The level of ability of the intended user relative to the resource discipline and educational level. This is closely related to Educational Context and Difficulty, but relates to the intended user, not the resource. Think of this as "Beginner, Intermediate, Expert" and their further refinements.

Security Level: This relates to an access control level, but does not include any mechanism for enforcement. For example, a resource in the US Department of Defense may be classified as "Secret". The management system may use this information for access control. Use of a standard classification scheme is strongly suggested.

Some concepts were considered granular enough, or without enough information available, that a single meta-data field was created. Educational/learningresourcetype was considered to have a well-defined vocabulary: its concept could be fulfilled with a single field. Definition and interpretation of this field is not considered limited or ambiguous. On the other hand, the field of educational/typicalagerange does not adequately address the concept of grade level. Grade level is clearly an important concept. If at some time in the future a single vocabulary of grade level emerges, it may be desirable to create a separate field of education/gradelevel as the field name itself would be so specific that it would not need to be defined via a structure. Until such a field exists, the classification structure may be used to enable description of the grade level that a resource addresses. The use of classification to describe educational grade level enables exploration of the concept. As the use results in converge on a single vocabulary or structure, a concept-specific field may be created as a permanent addition to the specification. A vocabulary is simply a taxonomy with only one level. Thus, one can include use of a defined vocabulary for a particular purpose by using the classification structure in the meta-data.

XML is a language of structure. There are several well-designed structures within the meta-data. The structures in meta-data that have already been defined include contribute and relation. They can be semantically extended without changing the structure itself. For example, contribute has a role that may have new values, effectively creating new types of contributors. But the structure of what constitutes information about a contribution is the same: a role, entities, and a date. The meta-data structures of contribute and relation create extensible definitions without the need to create new elements or fields. Extending the use of contribute does not require the use of classification. Structured information not adequately enabled elsewhere in the meta-data structure should be contained in the generic "structure" category known as classification.

The discipline (or subject) of an educational resource may be described using the classification category. There is a predefined classification purpose of "discipline".

	<purpose>
		<langstring lang="en">Discipline</langstring>
	</purpose>

<classification>
	<purpose>
		<langstring lang="en">Discipline</langstring>
	</purpose>
	<taxonpath>
		<source>Dewey
		</source>
		<taxon>
			<id>300
			</id>
			<entry>
				<langstring lang="en">Social Sciences</langstring>
			</entry>
		</taxon>
		<taxon>
			<id>320
			</id>
			<entry>
				<langstring lang="en">Political Science</langstring>
			</entry>
		</taxon>
	</taxonpath>
	<taxonpath>
		<source>LCSH
		</source>
		<taxon>
			<id>B
			</id>
			<entry>
				<langstring lang="en">
					Philosophy, Psychology, Religion
				</langstring>
			</entry>
		</taxon>
		<taxon>
			<id>F
			</id>
			<entry>
				<langstring lang="en">Psychology</langstring>
			</entry>
		</taxon>
		<taxon>
			<id>180
			</id>
			<entry>
				<langstring lang="en">
					Experimental Psychology
				</langstring>
			</entry>
		</taxon>
	</taxonpath>
	<description>
		<langstring lang="en-US">
			principles of operant conditioning
		</langstring>
	</description>
	<keywords>
		<langstring lang="en">operant conditioning</langstring>
		<langstring lang="en">psychology</langstring>
	</keywords>
</classification>

Note that the description and keywords relate specifically to the purpose of this classification, namely, "discipline". The combination of taxnonpath(s), description and keywords provides a rich definition of the discipline this resource addresses.

The keywords may all be aggregated upward into the general/keywords field to provide a complete set of keywords for the resource. The entrees from the classifications may also be included in the aggregated keywords. Imagine a search in which the user first inputs some keywords, gets a partial (called a "scan") and meta-data records back. All of the keywords might come back. The user then might ask for a sort by purpose, and get the keywords now sorted, with the short descriptions for embellishment and the taxonpaths presented in stepped (indented) form. This would allow a user to evaluate the resources quickly, perhaps refining the search.

A specific case of educational/difficulty will serve to illustrate the use of classification to elaborate an existing field. Let us presume that KnowledgePlanet.com (KPcom) has developed a system for rating difficulty according to several different potential user populations. For simplicity's sake, let's just say that at this point you have a rating by grade level. One can grind this ever finer, but this will suffice for illustration. You thus have a two level taxonomy of difficulty (purpose of educational difficulty in Classification):

Educational/difficulty:
   Grade
      Rating

For example,

Grade 1
      Rating 4 (very difficult)

The same item may also be rated for a different grade level:

Grade 3
      Rating 1 (easy)

Clearly an important part of the information that needs to be maintained with this set of ratings by grade is the source of the scale. This would be KPcom's scale.

In the meta-data classification this would have the appearance of (without going into XML):

Classification
   Purpose: educational/difficulty
   Taxonpath
      Source: KPcom Difficulty Scale A
      Taxon (1)
            Entry: Grade 1
      Taxon (2) 
            Entry: 4
   Taxonpath
      Source: KPcom Difficulty Scale A
      Taxon (1) 
            Entry: Grade 3
      Taxon (2) 
            Entry: 1

The Taxons within a Taxonpath are an ordered list of specific nodes in a path or stairway down through a taxonomy, starting at the top. Thus, one has created a set of difficulty ratings of a resource according to a specific source that includes ratings for several grade levels. I have left out the "ID" portion of the Taxons for simplicity. ID is optional, and would reflect the schema that KPcom used to define its taxonomic structure, if used at all.

One could use the ID as follows:

Classification
   Purpose: educational/difficulty
   Taxonpath
      Source: KPcom Difficulty Scale A
      Taxon (1)
            Entry: Grade 1
      Taxon (2) 
            ID: 4
            Entry: very difficult

At first one would expect to see a plethora of sources. Each organization may choose to have its own rating scale system. In time, those who specialize in Difficulty rating systems might start to develop cross mapping between rating systems. Or organizations may choose to label resources using several different rating scale systems by including more Taxonpaths. If the number of rating systems converge on one common rating system, then at some fuzzy time in the future, this rating system might be adopted by the entire community as the standard rating system to be used in educational/difficulty. We would hope that this might occur, but recognize that it would be impossible and inappropriate for the LOM to presume that it knows the best rating system. That is the domain of the community of practitioners.

The emerging strategy is to recognize that an important concept, such as educational difficulty, should not be left out of the meta-data. It may be included with a very coarse, poorly defined scale or vocabulary, e.g., difficulty: 0..4. Something is better than nothing. Further, the use of classification allows the inclusion of rating and vocabulary systems that are managed by specific entities (source) to provide an elaboration or alternative to educational difficulty. To be useful, then, classification should include purpose, taxonpath(s) and source(s). Each taxonpath can have only one source, but multiple taxonpaths within a single classification (i.e., single purpose) may be used.

Online education is still too much in its infancy to presume to create all necessary and adequate meta-data fields and vocabularies. Therefore, it is also possible to include concepts that are not included in the standard LOM. This is one of the intended purposes of Classification. By creating a new Purpose, one can create a new field concept (e.g., Price, Linguistic Complexity, and so forth) and use a "Sourced" vocabulary or taxonomy to populate one or more Taxonpaths. If a new concept comes into widespread adoption, first the Purpose term might be added to the suggested vocabulary of Purposes, and in time, there may come to be a new Meta-data field specifically for that purpose. The standard will evolve in a controlled manner based on practice.

The Classification category, which can be repeated many times, allows the generation of new concepts within the meta-data without the necessity of specifically including that concept in the standard. The data structure is always the same, so it poses few problems for the database administrator. Once the structure has been established to manage a Classification, it should be able to handle any Classification purpose. This is one of the values of using the "Purpose" as a property with a value, rather than having a specific field which is essentially a property value with a data value. The cost is slightly more complexity in the original database structure. The value is de facto extensibility of the meta-data without redesign of the database or standard.

The beauty of "Classification" is not that it will hold all of the answers, but that it provides a simple, rational method for groups to evolve and test concepts without significant impact on the existing standard (if and when that standard occurs). Note that LOM has taken this approach relative to rights management: we tried to figure out a rights structure, those knowledgeable in the field said it was not adequate. So we (as a collective decision) stripped it down to a bare minimum, awaiting the work of those skilled in the field to mature. We left a hole rather than create a bad job that would have to be replaced. We know it's not adequate, but in the larger scope of educational object meta-data, it will let us proceed.

Note--I have used the notation of a slash (/) to indicate that the second term is a child of the first in the structure, for example, educational/difficulty. The structure of the educational category has many terms beneath it, some with significant structures of their own. I have adopted the slash here as the dot (.) notation implies a straight linear model in which each field is named separately from the top down. The IMS (and IEEE LOM) meta-data is structured, hence the use of the different notation.

Many of the terms in this guide are defined in the Glossary.
 

Author:

Thomas D. Wason, Ph.D. (aka Dr. Tom)
Wason Consulting
wason@mindspring.com