INST > Clients > JISC

Representing frameworks of skill and competence for interoperability

Simon Grant 2007-06-11

This is a draft intended for improvement. Please send me your ideas which will be acknowledged if used. Part of the general concept owes a lot to the thinking of Scott Wilson: needless to say I take responsibility for mistakes and details: it would be great if you could help me correct mistakes!

1. Objectives

There are three clear objectives which a framework for skills or competences can fulfill.

  1. To provide human readable definitions and explanations to clarify just what the definitions are intended to mean.
  2. To provide unambiguous global identifiers so that anyone or any system can refer to the skills or competences without risk of the definition being mistaken for a different one.
  3. To provide machine processable information about the interrelationships between different skill or competence definitions at any level of granularity, and to related terms and resources.

The greatest benefits can be expected when all three of these objectives are addressed together; but significant benefits may come already from fulfilling the first two objectives alone. These two are easy to implement in themselves, and are a useful fallback if the third objective is not attained.

Skill or competence definitions by themselves are not generally intended to constitute criteria for admission or selection. Firstly, definitions can be the reference point for evidence provided by a candidate, so that it is clear what the evidence is referring to. Secondly, a skill definition can be used in conjunction with definitions of range, method of assessment, and assessment result, to provide a clear account of a person's tested capability. It is possible to combine definitions of skill, assessment and result into one, but this is not recommended, as it prevents factorisation, and may thus lead to an even greater plethora of definitions.

2. Previous approach superseded

There are two reasonably well-known standards for representing skill or competence definitions. One is the IEEE RCD (Reusable Competency Definitions), based on the earlier IMS RDCEO (Reusable Definitions of Competence and Educational Objective). Another is the HR-XML Competency Schema. In principle, either of these could be used to represent definitions using XML, with a DTD or Schema helping to ensure at least some syntactic conformance to the standard.

One problem comes with representing relationships between competences, and between them and other resources. In none of the existing mentioned specifications is there an obvious place for general-purpose relationships to be represented. An alternative way to represent definitions along with a predefined set of relationships between them is through some kind of taxonomy or structured vocabulary. However, such approaches predefine both the allowed relationships and what is entered about individual terms, and the question is then whether the predefined structures fit what is desired to be represented.

In practical or operational terms, the problem is that existing specifications are ill-adapted for representing much existing material which currently define competences. One clear example is UK National Occupational Standards (NOSs). NOSs have a basic common structure, but the detailed structure of the information represented differs. While it is always possible to work around particular cases, it appears increasingly awkward and unnatural to do so, the wider the range of cases presented. While it would be possible to define an information model and representational schema able to cover all NOSs, that approach simply leads to a proliferation of specific models for specific applications, which would not help interoperability.

Personally, I was investigating these approaches with the expectation that something could be made out of them, and it was only with reluctance that I had to admit that they were not going to work properly for the practical purposes of representing real, existing materials which define skills or competence. There does appear to be a need for a different kind of approach, and one such approach is described here.

3. Target approach using XHTML and RDFa

This new approach is based on the principle of having one http URL as the identifier bringing together all three objectives first specified above. Having unified functions in this way is likely to be inherently less problematic for maintenance, version control, search, etc.

Human readability is straightforwardly achieved by having the URL resolve to a normal human-readable XHTML file holding the material specific to the skill or competence being defined. Unlike approaches which rigidly predefine content, this allows the complete original material to be available directly through use only of that one URL (with the possible inclusion of ancillary image files, for example).

The URL itself satisfies the need for a global identifier. Uniqueness is another matter. In practice, it is possible that the same competence definition may be held in various places each with its own URL. What is needed in this case is to represent clearly the fact that they are identical. There is more than one way of achieving this, but the way proposed here is for each definition to hold within itself suitably marked URLs of identical ones, in such a way that it is possible automatically to follow such links and to establish a set of identical definitions without resorting to search or centralized registries.

On the basis of such an XHTML file, any interrelationships, including the fact of being identical, can be represented unambiguously by embedding RDF information, using the RDFa approach. For example, this may represent the relationships between a NOS unit and its constituent statements; or between a higher level learning outcome and the lower level outcomes that are parts of the higher level outcome. There are many other relationships that people may want to represent in this way. A sensible approach would be to embed such information using RDFa as and when a potential machine processing use for it emerges which justifies the effort of embedding. To achieve interoperability, the terms used to designate the kind of relationships would need to be agreed. These terms in RDF are known as "predicates", and many suitable ones are already defined in W3C and DCMI materials.

Beyond representing the fact that two definitions are actually the same thing, in the context of learning, education and training there is a clear potential use for defining a very small number of extra predicates. The case for these that follow has been made elsewhere.

Operational equivalence:
Two definitions may have been made by different people using different words, and thus not be identical, but each authority may recognise that the other is equivalent for practical purposes.
Operational non-equivalence:
It may be useful to be able to deny operational equivalence.
Satisfaction:
If two definitions are not equivalent, it may be nevertheless that one of them is greater, more extensive, or with more coverage. This definition can be recorded as satisfying the less stringent or lower requirement. This relationship should be transitive.
Contribution:
Where one competence is satisfied by another, the first may or may not contribute materially to the second. This relationship could be seen as a variant of a "part-of" relationship. But requiring this to be transitive may not be useful.
Overlap:
Two definitions may be related, but none of the above relationships apply. In this case it could be noted that the definitions overlap. In many cases, it would be expected that two overlapping definitions could be broken down into constituent parts that would indeed relate in one or more of the above ways.

The advantages of this approach can be summarised as follows:

There are other approaches not reliant on RCD, RDCEO or HR-XML competency definitions, and detailed comparisons need to be carried out. The more generally applicable an approach is, and the less extra software it requires, the more desirable it would appear, particularly for independent learners.

4. Intermediate approaches

RDFa is currently a W3C draft, not yet a recommendation. A reading of published W3C HTML Working Group charter suggests that there is likely to be a full recommendation, along with a new version of XHTML, in 2010. There are sufficient tools and materials for investigatory and prototyping work to begin immediately, but if a solid basis is required in the meanwhile, compatible approaches can be explored that deliver most of the advantages envisaged above.

4.2 Using XHTML for the resources and RDF for relationships separately

In the favoured approach above, RDFa markup is added to XHTML to allow humans to read and machines to process the same file. One could stop short of this, maintaining the distinction between a human readable XHTML file and RDF files to express the relationships. A single link could be provided in the XHTML to the related RDF, with conventions that are already used, such as a <link> tag or a <meta> tag in the document head. If this were done, the identifier for the competence could still be the URL of the human-readable file. The related RDF file could contain exactly the RDF that would have been able to be extracted from an RDFa XHTML file, and could also express an RDF triple giving the URL of the human-readable XHTML.

The alternative, of giving the URL of the RDF as the identifier, is less attractive, as it would not be clear to a human where the human-readable information was.

4.2 Using any other XML for the resources

Defining a skill or competence in human-readable terms using XML is not confined to XHTML. It can also in principle be done using any XML format with an appropriate style sheet. Though, as noted above, existing specifications are not exactly suitable for much existing material, there is no reason why more fitting specifications using XML should not be devised.

If a different XML format is used, a reasonable approach might go about it in this way.

A minimalist variant of this would be to use the RDCEO/RCD approach for the XML, while explicitly excluding the use of the "definition" element. What remains is very similar to a generic blog entry.

5. Methodology

The first overriding reality to note is that the main challenges to be surmounted will be political ones. In particular, vital questions will relate to how definitions are set up, controlled and maintained in the first place, and then who takes the technical responsibility for establishing the domains and URLs, making the files or equivalent web services available, and ensuring that changes in definitions are reflected in the materials published on the Web. There are many examples of existing organisations that establish and maintain definitions themselves, but very many fewer that have established domains and URLs serving these functions. There is also the standardisation question of which body will be in a position to establish the effective standards that are needed.

A plausible method of achieving the objectives may include the steps listed in these two following subsections.

5.1 Steps for organisations wishing to define skills or competencies in this way

5.2 Steps for community or standards organisations

Other references

See previous work by me in this area.

page maintained by and © Simon Grant, edition 2007-06-11