Chapter 7
TARGETING LEARNING RESOURCES IN
COMPETENCY BASED ORGANIZATIONS: A
SEMANTIC WEB-BASED APPROACH
Anne Monceaux1, Ambjörn Naeve2, Miguel-Angel Sicilia3, Elena GarciaBarriocanal3, Sinuhé Arroyo3 and Joanna Guss4
1
EADS-Innovation Works,18, rue Marius TERCE, Toulouse, France –
anne.monceaux@eads.net
2
Royal Institute of Technology (KTH), Stockholm & University of Uppsala. Uppsala Learning
Lab (ULL), Kyrkogårdsgatan 2 C Uppsala, Sweden;– amb@nada.kth.se
3
University of Alcalá de Henares. Polytechnical Building (Computer Science Dept.).Ctra.
Barcelona km. 33,600. Alcalá de Henares, Spain – msicilia@uah.es , elena.garciab@uah.es,
sinuhe.arroyo@alu.uah.es
4
EADS France – Innovation Works, Learning Systems - CTO-IW-SE-LE, 12, Rue Pasteur BP76 - 92150 Suresnes Cedex, France – joanna.guss@eads.net
1.
MOTIVATION AND PROBLEM DESCRIPTION
Recent standardization and specification efforts in the area of learning
technology (Friesen, 2005) have resulted in a considerable improvement in
the interoperability of learning resources across different Learning
Management Systems (LMS) and Learning Object Repositories (LOR).
Examples are the ADL SCORM and IMS Learning Design specifications,
which provide shared languages to express the packaging of learning
contents and learning activity designs respectively, among other elements.
The central paradigm of such reuse-oriented technology is the notion of
learning objects (LO) as digital reusable pieces of learning activities or
contents. This represents an opportunity for organizations to devise more
effective mechanisms for targeting learning activities internally as a way of
improving their capacity to respond to the changing business and
technological environments and also to the evolving customer needs.
However, transportability of digital learning objects across platforms is
only a basic step towards higher levels of automation and possibilities of
delegation of tasks to software agents or modules. Such advanced
technology requires richer semantics than those offered by current metadata
specifications for learning resources (Sicilia and García-Barriocanal, 2005).
Semantic Web technology and the use of ontologies are able to provide the
required computational semantics for the automation of tasks related to
learning objects as selection or composition. In general, they enable new
possibilities to enhance organizational learning or even fostering systemic
learning behavior inside the organization (Sicilia and Lytras, 2005). In
addition, Semantic Web Services (SWS) provide the technical architecture
and mediation facilities for semantic interoperability required for selection
and composition of learning objects in a distributed environment in which
there are potentially many heterogeneous repositories (Lama et al., 2006).
Within the context described, the dynamic search, interchange and
delivery of learning objects within a service-oriented context represent a
major challenge that needs to be properly addressed. In short, this entails the
technical description of the solution in terms of SWS technology, and also
the provision of the ontologies, facilities and components required to extend
and enhance existing learning technology systems with the advanced
capabilities provided by computational semantics. Semantic Web Services
provide the required conceptual representations, along with the capabilities
to translate and integrate diverse systems that share the common goal of
reusing learning objects. A Semantic Web Service engine integrated with
existing standardized LMS technology will extend the possibilities of
learners, tutors and instructional designers with semantic search tools
capable of asking for and retrieving learning objects from any provider that
registers itself as a Semantic LOR. Semantic Web Services, as conceived in
the WSMO framework29 provide the required ontology-based representation
flexible enough to specify realistic learning needs and exploit domain or
specialized knowledge in the process of search for learning objects (Lama et
al. 2006). A key feature of WSMO is the ontological role separation between
user/customer (goal) and Web Service. This matches the concept of learning
tasks being separate concept in learning literature. However, before a SWS
architecture can be fully exploited, there is a need to devise the underlying
framework for the expression of learning needs and their subsequent use for
selecting learning resources. This chapter addresses one concrete way of
expressing such learning needs in terms of competencies, which are
especially adequate for organizational learning.
Competencies have been defined in terms of observable human
performance, (Rothwell and Kazanas, 1992) encompassing several elements:
29
http://www.wsmo.org/
(1) the work situation is the origin of the requirement for action that puts the
competency into play, (2) the individual’s required attributes (knowledge,
skills, attitudes) in order to be able to act in the work situation, (3) the
response which is the action itself, and (4) the consequences or outcomes,
which are the results of the action, and which determine if the standard
performance has been met. This kind of definitions leads to a paradigm of
competency computation in which both organizational needs and the
expected outcomes of learning resources are expressed in terms of
competencies, thus enabling numerical or symbolic accounts of the
competency gap, i.e. the (amount of) competencies that are required to fulfil
some give needs or to reach a more desirable status in organizational terms.
Existing work on engineering competency ontologies (Sicilia, 2005) has
resulted in flexible models that can be used for the critical task of targeting
learning activities inside the organization, personalized to the competency
record of each employee. This chapter reports on the early implementation
of such approach in a concrete organizational context.
With the aim of exploiting the advantages of a Semantic Web Service
Architecture to make richer and more flexible the processes of query and
specification of learning needs in the context of Learning Management
Systems and Learning Object Repositories, a use case centered on
competency-based selection in the Aeronautic field is depicted in the
following. Based on this analysis, the viability and benefits of the approach
are presented and briefly discussed at the end of the chapter.
2.
COMPETENCY DRIVEN TRAINING SELECTION
IN THE AERONAUTICAL FIELD
Training significantly contributes to the companies’ ability to react on
requirements of fast changes markets, customer needs and successful
business process. Nowadays, aeronautical industries have a high demand for
well-trained teams. At the same time they face continuous changes in their
work processes and tools. Not only is continuous education an important
process but it is managed on a contractual basis. Therefore, training
management activity is a common responsibility of Human Resources
(HHRR) departments. Actions and decisions about training are taken by
HHRR according to the company objectives. The important requirement for
training management is that it supports developing and maintaining the right
range of skills and competencies needed for the employees’ jobs.
The present use case aims at improving the way in which Training
Management can work towards this goal. More in detail, on how to better
mediate among domains by reusing or integrating knowledge that results
from competency management activities for training selection. Thus, it can
be stated that the ultimate mission of Training management is to support
Competency management.
In the following, a brief depiction of the main aspects of the use case is
presented. Such depiction helps elaborating about the benefits of a Semantic
Web-based approach to e-Learning, same for the particular Aeronautical
scenario presented as for learning activities in general.
2.1
Actors and roles
Several actors participate in training management processes:
1. Training Manager from the Human Resource Training Department. The
Training Manager takes responsibility for managing training plans
according to the business strategy, as well as training budgets and
requests.
2. Employees, including Engineers and their Team Managers. They are the
originators of training requests.
3. Training organisms provide the training offer, including training
materials and courses.
Figure 7-1. Main training processes
In the following section the particularities of the use case are briefly
depicted. Figure 7-1 provides an overall view of the main elements in the
training management process. The concept of competency can be used in
such processes as the language for expressing needs, match learning
resources, record the employee profile and measure the effectiveness of the
training activities.
2.2
Training related information objects
The HR Training Department activity uses and produces various training
related information. In the following, the different categories of information
objects of interest for the use case are presented.
Structured Training Packages. HR manages some LO references and
description in an SAP database. The granularity level under consideration
is that of Structured Training Packages (Naeve et al 2005).
Core training catalogue. Metadata elements are used for publishing
purposes. Web training catalogues are rendered accessible through the
various subsidiary companies’ intranets. These training catalogues are
online abstracts of the real SAP database.
People training history. SAP database also allows the management of the
people Training History. Human Resources keep track of requested,
planned, rejected, accepted or completed training sessions for every
Employee. Thus, it is possible to know about the training sessions
followed by a given Employee or about the status of a given Training
Request.
All these materials are currently stored in databases and independently
maintained. A topic hierarchy (See Figure 7-2) is used to filter accesses to
the SAP training database. By this means, specialized training engineers
benefit for an accurate information access and are made responsible for such
or such topics. It is also used to structure the display of the web training
catalogue on the intranet.
Figure 7-2. Training Topic thesaurus (excerpt)
Training packages can be assimilated to a very specific kind of learning
objects, and in consequence, they can be annotated with the competencies
that are the expected outcomes of the training. This is already considered in
the IEEE LOM learning object metadata standard (in which classifications of
learning objects may state competencies) and thus provide room for
describing the resources in terms of complex models or ontologies of
competencies.
2.3
Training requests
The employees’ training requests are addressed to the Training Manager.
Two different cases are considered when it comes to deal with training
requests:
Individual training request: It involves an Engineer who wants to follow a
particular training or express ‘informal’ need.
Competency driven training request: The request is a result of an annual
interview between an Engineer and her/his Team Manager. This
interview results in the Engineer competency profile update, and an
agreement on associated training needs.
The training history serves as a way to encode the competency record of
employees, so that competency driven requests can use that information as
input. This step can be supported by the use of ontologies of competencies,
which can be used to “suggest” possible paths of competency acquisition
and the associated resources/activities that could be used in each of them.
After the eventual completion of the activities, their effectiveness can be
used as input for future learning activities, thus closing the loop.
2.4
Competency index and profiles
Competency management, although in the sphere of Human Resources, is a
parallel process generating its own information flow and data. The
categories of information objects related to competency management are
presented below:
Reference competency index: Lists competencies, skills and knowledge
involved in professions needed by the organization;
Position profile: Resorts to the reference index to define scaled required
competencies and skills at a given position. Several positions may come
under the scope of a same reference profession, while requiring different
proficiency levels.
Personal profile: Resorts to the reference index to define scaled actual
competencies and skills of a person holding the given position.
The use case intends to make them reusable for training retrieval and
selection, allowing the calculation of a competency gap between a target
position profile and an employee profile. Thus, job positions serve as
stereotyped models of competency aggregations.
2.5
Viability and benefits
As shown above, various data and systems are involved in answering
training requests taking in account the needed and available competencies.
Resorting to Semantic Web-services for the selection and combination of
training courses requires that:
The training search function supports selection/combination and allows
taking a competency gap description as criteria. This means:
Handling queries with various concepts (competencies, professions,
topics, etc.) from separated data sources: training database or other
LCMS, (training history), profession competency index, etc.
Handling position profiles and employee profiles to build competency
gaps.
Handling competency profiles and using them as criteria for selecting
trainings.
Handling LO target competency or pre-requisite and use them as criteria
for combining trainings.
LO-based training descriptions include competencies. The key point towards
context-aware learning object delivery in the aeronautical context is that
both, trainings goals and pre-requisites must be described in terms of
competencies. This is where a different problem occurs, related to the
cost of manual annotation in time and resources, especially when the
training database is continuously evolving to reflect updated offers.
A unified model applicable to the Training management and Competency
management domains supports the indexing of Training Packages using
Profession / competency referential; and the retrieval / selection /
combination services over Training Packages.
Ontologies of competencies (Monceaux and Guss 2006) provide a rich
description framework for the selection of resources, which can be extended
with a organizational process view as that described in (Naeve and Sicilia
2006). The benefits in terms of increased decision support are evident from
the above, and the organization could also benefit from the systematic
approach to defining competencies required. However, an assessment of the
viability also requires reflection on the technological challenges required.
These entail the storage of competency databases and the development of
query resolvers that handle the abovementioned elements. The results of the
LUIS project provide the framework for these issues. In consequence,
organizations that do actually have a “competency culture” can benefit from
semantic technology directly, since the requirements on management and
recording of competencies are currently covered by non-semantic
technology, perhaps with the exception of some practices as the formal
annotation of learning resources with a statement of the competencies they
are intended to provide. This is thus a case of technological enhancement on
existing practices.
3.
THE OVERALL PROCESS VIEW: A
COMPETENCY GAP APPROACH
In a service-oriented environment that aims for reusability of service
components, the “process-object” – or “process-module” is of vital
importance. In this section we will discuss how such process modules can be
used as contextual units, e.g., connecting learning objects with learning
objectives and competency gaps. Moreover, we will show how such process
modules can be connected into service networks, whose overall service goals
can be seen as aggregated from and composed of the sub-goals of the
participating process modules.
3.1
The Astrakan™ process modelling technique
The basic ideas underlying the Astrakan™ process modelling technique30 are
depicted in Figure 7-3.
30
www.astrakan.se
Figure 7-3. The Astrakan process modelling technique
A Process Module has certain Process Goals, produces Output Resources
for different Stakeholders, refines Input Resources and makes use of
Supporting Resources (Figure 7-4). The difference between an input- and a
supporting resource is that the former is refined in the process, while the
latter facilitates this refinement.
Figure 7-4. A Process Module with its Goals, and its Input-, Output-, and Supporting
Resources
Figure 7-5 depicts a kind of (= subclass of) Process Module, called a
Learning Process Module (LPM) with its corresponding Learning (Process)
Goals, and its Input-, Output-, and Supporting Learning Resources.
Figure 7-5. A Learning Process Module with its Learning Goals, and its Input-, Output-, and
Supporting Learning Resources
Observe that, in Figure 7-5, the Learning Process Module (LPMs) provides
the crucial connections between Learning Resources (LRs), which include so
called Learning Objects (LOs)31, and Learning Goals (LGs). Hence, it
becomes possible to describe why we are using certain LO in a particular
LPM, i.e. what pedagogical aspects that we are trying to support and what
LGs that we are trying to achieve. Apart from the never-ending debate about
their definition, a major criticism against LOs is that they are too often
considered in isolation from the learning context within which they are
supposed to be used. Hence it becomes difficult to connect LOs with the
social and pedagogical dimensions of the learning process, and answer the
crucial pedagogical/didactical questions of why LOs are being used and
what one is trying to achieve by using them. By applying the modeling
techniques introduced in (Naeve et al. 2005) and elaborated in (Naeve and
Sicilia 2006), such questions can be answered in a satisfactory way.
3.2
Different types of Competency Gaps
Since individual competencies are refined and developed by learning, they
can be considered as input and output data to learning processes. In fact,
each Learning Process Module (LPM) can be considered as filling a Real
Competency Gap (RCG), which is the difference between the Input
Competency (IC), i.e., what the learner knows before entering the LPM, and
the Output Competency (OC), i.e., what (s)he knows after having passed
through it. The Formal Competency Gap (FCG) is the difference (as
specified e.g., in a course manual) between the Pre-Requisite Competency
(PreRC), which is required to enter the LPM, and the Post-Requisite
Competency (PostRC), which is the competency that the LPM aims to
provide for learners that fulfill its corresponding PreRC.
In Figure 7-6, the ICs and OCs are modeled as a kind of Learning
Resources, while PreRCs and PostRCs are modeled as a kind of Learning
Goals. Pre-assessment can be used to investigate whether there is a Pre
Competency Gap (PreCG), i.e. whether there is a difference between what a
learner knows when entering the LPM, and what (s)he should have known in
order to enter it. Post-assessment can be used to investigate if the learner has
actually acquired the aspired PostRC. If not, then there is a Post Competency
Gap (PostCG), i.e., there is a difference between the PostRC and the actual
OC for this learner. If there was no PreCG, then we can conclude that
something went wrong in this LPM.32
31
As well as other types of resources, such as human resources and physical resources
(materials, tools, laboratories, etc.)
32
This is analogous to a software principle called “design-by-contract”, where only data that satisfies the
pre-conditions are allowed to enter a software module. If the post-conditions are not fulfilled, then we
can conclude that something went wrong in this module.
Figure 7-6. A Learning Process Module with a Formal and a Real Competency Gap
A Forward Competency Gap (FCG) is a difference between what the learner
knows and what (s)he plans to know, while a Backward Competency Gap
(BCG) is a difference between what the learner knows and what (s)he should
have known. Hence, with respect to an LPM, a BCG is identical to a PreCG.
In the EADS use case, the difference between an employee’s Personal
Profile and her/his Present Position Profile is her/his BCG. The difference
between the employee’s Personal Profile and her/his Desired Position
Profile is her/his FCG.
In general, FCGs are more associated with strategic learning needs (what
a company needs to learn in order to stay in business), while BCGs are more
associated with operational learning needs (what a company needs to know
in order to deliver in its present undertakings). BCGs often appear because
employees leave the company and have to be replaced by others who do not
quite know what they (ideally) should have known in order to serve as good
replacements.
3.3
Competencies as Connectors of Learning Process
Modules
A Learning Process (LP) can be modelled as a chain of successive LPMs,
where the PostRC of the LPMk is identified with the PreRC of the LPMk+1.
In this way, the large learning goal of the entire LP can be broken down into
a sequence of smaller learning (sub)goals for each LPM. This map well to
the concepts of goals and sub-goals in WSMO, where there are ggmediators, use to meditate mediate between goals.
Figure 7-7. The EADS employee competency model
3.4
Modeling with a general competency ontology
Reuse-oriented learning technology emphasizes the role of metadata that
describes the properties of learning resources as a mean to provide advanced
support for the location and selection of learning resources. These properties
are of a various kind, but one of its principal categories is that of describing
the learning needs the resource facilitates in some way. In semantic
approaches to learning technology, ontologies that enable the description of
learning needs are thus a critical piece. Learning needs can be stated in many
different ways and can be considered to be dependant on theories of learning
to some extent. Among them, the concept of competency emphasizes the
specification of external, observable behavior oriented to performance in
activities. In organizational contexts, this entails that competencies are
oriented to describe performance in concrete work situations.
The literature on formalizing competencies to date is scarce and
fragmentary, and specifications dealing with competencies as HrXMLCompetencies33 or RCDEO34, while useful for data interchange, do not
provide the required computational semantics. A general purpose schema
for competencies (call GCO –General Competency Ontology –) based on the
schema describe in (Sicilia, 2005) has been approached in an attempt to
increase the re-usability and flexibility of the resulting technologies.
3.5
Addressing flexibility in the definition of the
competency concept
Flexibility in competency specification is currently approached in the
ontology in two ways. On the one hand, a competency definition is made up
of competency elements, and competency elements are specialized in several
components (skills, attitudes and knowledge elements in actual version),
allowing for the inclusion of other elements in the future. On the other hand,
current schema allow for incomplete definitions of competencies. A
competency is completely defined if it is explicitly indicated as such, and
this entails that the presence for an individual of all the elements that
compose the competency is a necessary and sufficient condition to describe
the competency. A competency can be partially defined if it is defined as a
primitive competency (i.e. its elements are not defined) or if the described
components do not define the competency completely.
See the following example (Figure 7-9): The competency “Programming
Java with Eclipse” is composed by two knowledge elements “To know
Eclipse environment” and “Programming Java”. The competency has been
explicitly defined as a completely defined competency. If a person P1 has
acquired both knowledge elements, a reasoner can deduce that this person
has the competency “Programming Java with Eclipse”, although it is not
expressly stated.
The general competency model described in this section is used in the
architecture of the LUISA project35. Figure 7-8 depicts an scenario inside
LUISA, in which a search component talks to the Negotiation Layer (a part
of the SWS infrastructure) to get matches for some given competency gap.
The resources are stored in (one or several) LOMR (learning object metadata
33
http://www.hr-xml.org
http://www.imsproject.org/competencies/
35
http://www.luisa-project.eu
34
repositories), and the metadata in such repositories can be edited through
SHAME tools36.
Figure 7-8. Scenario from the LUISA architecture
Figure 7-9. Example of completely defined competency
3.6
Competency Components
One important issue to deal with in the ontology refers to the need of
separate actual competencies, associated to particular individuals, and the
definition of competencies as stereotypes. Given that the Competency
concept represents a discrete competency of an individual generally
36
http://kmr.nada.kth.se/shame/
portrayed as processors. Such processor provides room for software systems
that are able to exhibit some competencies.
On top of that, the elements influencing competencies are of a various
kinds, including knowledge, skills, abilities, and also attitudes. By using
these concepts a clear separation about three types of traits that represent
different aspects of competency is clearly achieved.
For example, an employee may have the knowledge about the different
phases of a given internal process, since he or she has attended trainings
about it. This is different than having the skill of implementing the process
correctly. In fact, the knowledge about the internals of the process may not
be necessary for its proper usage, and on the contrary, knowing the internals
does not guarantee that the employee is able to use the process efficiently. In
addition to that, attitudes represent elements that are not necessarily
connected to specific knowledge or skills. For example, having good
influencing skills does not always entail that an employee would have the
attitude to make his/her opinion prevail. Figure 7-7 provides a screenshot of
the modeling of EADS competency ontology. In that case, the terminology
was slightly different, but after a mapping phase, they were assimilated to
similar concepts in the GCO.
It should be noted that from an ontological perspective, attitudes are
mostly domain independent, while knowledge items and skills are not.
Examples are “service orientation” or “attentive to details” attitudes that are
equally applicable to employees, irrespective of the industry. Some skills are
also of a generic nature, like “social aptitude” or “leadership,” but many
others refer to concrete elements or artifacts that are specific of the industry.
Typical examples are “PHP programming skill,” “Unix administration,”
“repairing Aston Martin engines,” and the like.
The part of the current version of the ontology that models competencies
and competencies definitions is depicted in Figure 7-10. For the sake of
clarity, not all the ontology properties are shown.
Figure 7-10. Partial graphical view of the ontology: Competencies and Competencies
definition.
3.7
Work situations
Competencies are put into play in concrete job situations, which can be
considered as a kind of Episode in the life of the organization that occurs at a
concrete moment in time. The consequence attribute in the concept
JobSituation simply represents the outcome of the episode, which can be
used as a source of assessment for various purposes, including the revision
of the beliefs the system has about the competencies of the participants.
Competencies and job situations are connected to their respective
“definition” elements. These definitions are used to represent stereotypical
competencies and job contexts, so that they can be used to describe, for
example, job position characterizations in human resource selection
processes, or as a way to state the needs of a project.
Each job situation definition requires a number of competencies as
defined in CompetencyDefinitions. This is a way to describe work situations
in terms of required competencies.
Figure 7-11 briefly depicts work situations in the current version of the
ontology.
Figure 7-11. Partial graphical view of the ontology: Work situations
3.8
Relationships between Competency Specifications
Competency specifications are implicitly related by the relationships among
competency components. For example, if a competency c1 is considered to
require some knowledge k1 then, the competency implicitly requires the
knowledge of any k1 pre-requisite knowledge. This is represented through
the prerequisite relationship (knowledge trees can be modeled this way).
Skills can also have knowledge elements as prerequisites, and they could be
considered to be composite (not in that version of the ontology).
Relationships between competencies can be of a diverse kind. Initially,
we only deal with prerequisite and details relationships here. The latter is
conceived as a form of “specialization” in the sense that a competency
provides a more detailed description to an existing one. For example,
“Administering Oracle databases in large installations” stays at a higher
degree of abstraction than “Administering Oracle 9.0 databases in large
installations.” The specialized competency usually requires more specific
knowledge elements. Both the “prerequisite” and “details” relationships
entail some form of prerequisition, but the semantics are not exactly the
same. For example, the C1 ≡ “ relat ional dat abase design”
competency is a prerequisite for C2 ≡ “ Adm inist ering dist ribut ed
Oracle databases in large installations”, but it is not a detail, since it
reflects only a previous component of knowledge. In other words, the
competency C2 cannot be considered as a specific kind of competency C1.
Some other simple competency relationships are equalTo and similarTo.
The former is a simple way to state that two competencies are the same,
while the latter is a way to express different strengths of correlation or
resemblance between competencies.
Figure 7-12 depicts relationships between competencies in the current
version of the ontology.
Figure 7-12. Partial graphical view of the ontology: Relationships between competencies
3.9
Defining Competency measurement scales
Measurement scales for competencies can also be of a diverse nature.
Although the development of simple integer scales is common, other kind of
scales could also be allowed. In the ontology, a Measurement is connected to
competencies as an elaboration of the simple Level attribute of the
Competency concept in Figure 7-12. Measurements are always related to a
given MeasurementScale, and usually some MeasurementInstruments
associated to such scales are available (e.g., questionnaires or interviews).
From this basic level, several types of scales and their associated
measurements can be defined. Specific scales can be defined as an instance
of IntegerMeasurementScale. Each scale must provide some definitions that
act as constraints on the description of the measurements.
In the ontology, a JobPosition is described in terms of competency
definitions by specifying a given MeasurementLevel, connected to the scale
in which the level is expressed. This is an example of how other elements
different from processors can be described using the ontology. The elements
in current version of the ontology could be complemented with other
ontology terms that better describe each measurement instrument, and also
with “conversions” from one scale to another, when available.
Figure 7-13 depicts the part of the current ontology that represents
measures. For the sake of clarity, properties have not been labeled.
4.
ARCHITECTURAL SOLUTION
In the following the architectural solution provided is putting in the context
of the use of competencies. Firstly, the LUISA architecture is presented.
Secondly, how it is applied to computing competency gaps is sketched.
Finally, the particularities on how it tackles the use of training resources by
means of a specialized QueryResolver are carefully depicted.
4.1
The LUISA Architecture
Currently, the LUISA includes the following core building blocks that are
directly related to the competency approach:
Competency Gap Search: It provides the means for finding competencies
and filling the competency gap. In short, given a target position and the
employee profile, the competency gap search takes care of calculating
the competency gap, requesting LO using a specific negotiation layer
protocol and finding an appropriate set of learning objects that can in
principle fill the competency gap.
Negotiation Layer: The negotiation layer fulfils a two-fold purpose. It
receives requests for LO expressed using the specific negotiation
protocol. It also takes care of providing the results of a particular
competency gap search. This layer is an interface for the WSMO-based
SWS layer that integrates heterogeneous sources of LO by means of
semantic description of conventional Web Services.
SHAME: Taking as input the results of a competency gap search the
SHAME plug-in takes care of presenting them as appropriate. In order to
obtain all the necessary metadata it closely communicates with the
LOMR module.
LOMR: The LOMR receives LOM metadata from the SHAME plug-in.
Additionally, it provides SHAME with RDF/XML blobs, also using the
same annotation protocol for its presentation.
Figure 7-13. Partial graphical view of the ontology: Measurements in ontologies
4.1.1 Computing competency gaps
Competency gaps are calculated by subtracting the training requirements
from a particular target position from trainings already attended by a given
employee as detailed on his profile. As a result a collection of metadata that
needs to be mapped to learning objects descriptions is produced. Once such
process has been achieved, it is the task of the negotiation layer to locate the
most appropriate set of learning objects for filling the competency gap.
Should that matching set of LOs be available, the competency gap, is indeed
filled. Of course this is only one of the many possible competency gap
analyzers that could be devised, but it serves as the ground for the future
integration of other, perhaps more complex, analysis schemes.
4.1.2 Targeting training resources through an specialized
QueryResolver
One of the main tasks of the specialized QueryResolver is to manage queries
that affect multiple data sources. In short, it takes care of consolidating the
results obtained from training repositories, training histories, LCMS or any
other that needs to be checked in order to bridge a competency gap.
Additionally, it provides indexing capabilities for easing and speeding the
combination and location of materials. The idea of having a QueryResolver
for gap analysis enables the design of several of these components that could
be “pluggable” inside the LUISA architecture.
5.
CONCLUSIONS
Competencies represent a paradigm of observable workplace behaviour that
can be represented in terms of ontologies. These ontologies can be used for
the expression of learning needs, and also for the expression of the expected
outcomes of these activities. This creates a link between needs and resources
that can be exploited for advanced targeting capabilities. A concrete case of
organizational learning has been described, followed by a general conceptual
model that details how competencies and learning resources can be mapped
in a process-oriented framework. Competencies provide the organizational
meaning to learning resources, and they can be used as input and outputs in
learning process models. Finally, the results of project LUISA have been
described as a technical solution using Semantic Web Service technology
that uses a given, flexible and generic competency schema. The LUISA
solution provides the required semantic technology to fulfil the needs of
competency-centric approaches to organizational learning of any arbitrary
complexity.
6.
QUESTIONS FOR DISCUSSION
Beginner:
1. Competencies are a model of observable human behavior that is widely
used in the literature about organizational learning. However, the term
“competency” (plural competencies) is used in the literature with
different meanings. The GCO model presented in this paper provides a
flexible definition of competencies but other ways of referring to the
same things can be found. The following questions are oriented to:
It is common in the literature on theories of learning to refer to concepts
as declarative knowledge, procedural knowledge and values as related
to learning. How these three terms relate to the competency elements
skills, knowledge items and attitudes?
The O*Net database37, containing information on hundreds of
standardized and occupation-specific descriptors. They include the
concept of “skill”. How does the O*Net concept of skill related to the
GCO described in the chapter? And how the rest of the concepts in
O*Net map to the GCO?
2. The key ingredient of competency-based approaches is a correct
understanding of what competencies, their components and their
relationships are. This is an understanding and analysis phase that
requires reflection on how competencies are defined and measured inside
the organization.
Intermediate:
1. Expressing organizational needs in terms of competencies requires some
kind of forecasting or at least a consideration of the requirements for the
short terms regarding the capacity of the employees. The competency
gap then expresses the competencies (or competency components)
currently not available among the employees, and then the process of
matching and targeting learning resources (learning objects) takes the
gap and attempts to select the best learning activities/contents that can be
used to facilitate the learning process that eventually might result in the
required increased human capacity. However, this process is not as
simple at seems at first glance, and many issues that require the use of
complex models – as those that can be expressed in terms of ontologies –
demand attention. The following questions are oriented to reflect on
some of these issues:
How can learning resources be described to facilitate search in terms of
competencies?
Once learning activities have been programmed and carried out by the
target employees, there is a need to evaluate the acquisition of the
required knowledge. How does this impact the assessment of the
learning resources for future learning programs?
How can the agenda and constraints of the employees be taken into
account in the delivery of the learning activities resulting from a
competency gap analysis process?
2. Learning object metadata in semantic form is an alternative for resolving
question (1) – some answers can be found in (Sicilia, 2006). Question (2)
points out to the possibility of using the evaluation of the activities to rate
37
http://www.onetcenter.org/
in some way the learning resources used, so that those that have
effectively facilitated the required learning are considered best, and those
that have not can be considered to be discarded or improved. Question
(3) introduces the complex issues of time planning. Gap analysis can be
combined with temporal (or spatial) constraints for a more informed way
of targeting learning activities inside an organization.
Advanced:
1. There is not a single, universal approach for computing competency gaps
from a given record of competencies. This is among other factors
because the relationships between competencies can be of very different
natures; there is not a universal method to assess competencies and
competencies can be described at different levels of granularity. In
consequence, competency ontologies as the GCO described in the
chapter are “upper models” or general schemas that can be used and
extended in several directions. The following questions pose some of the
issues that could be considered for concrete applications.
How can competencies be measured for particular employees? Since
competencies are related to workplace performance, which type of
methods are more reliable? Peer assessment might be one of them?
How can competency components be aggregated into composite
competencies? What is the difference between dependencies between
competencies and competency components? What are the
implications of these kind of issues for computing competency gaps?
When considering a concrete organizational need expressed in terms of
competencies, the matching process should require an exact mix of
competencies? In other words, if some employee possess competency
level 3 for competency X and the requirement is a level of four, could
this be compensated, for example, by an “excess” in other of the
required competencies?
All these questions are actually research questions and they do not have a
unique answer. Many different approaches can be devised considering
variants of the algorithms of gap analysis and/or tailored models of
competencies. This is essentially the approach of the LUISA project,
different QueryResolvers can be used to implement different (perhaps
competing) approaches, creating opportunities for contrast and
customization.
7.
SUGGESTED READINGS
The recent book on competencies in organizational e-learning edited by
Sicilia (2006) provides a selection of chapters about the competency
approach for organizational learning. It includes chapters on the key
organizational dimension, but also several chapters that describe concrete
applications of Semantic Web technologies to managing competencies.
As such, it is an excellent complement to the approach described in this
chapter.
The description of learning resources can be accomplished through
metadata. Standards and specifications regarding different aspects of
learning-oriented metadata are introduced in Friesen (2005). After a basic
understanding of the specifications mentioned by Friesen is achieves, it is
worthwhile to go through some papers that deal with the extension of
such standards with Semantic Web technology. Many examples can be
found in the “Applications of Semantic Web technologies for eLearning” (SW-EL) workshop series38, and Sicilia and GarcíaBarriocanal (2005) can be used for a general understanding of the issues
behind those approaches.
8.
ACKNOWLEDGEMENTS
This work has been supported by project LUISA (Learning Content
Management System Using Innovative Semantic Web Services Architecture),
code FP6−2004−IST−4 027149 and by project SERE-OC (Resource
selection in e-learning based on competency ontologies), code CCG06UAH/TIC-0606 funded by UAH and the CAM (Comunidad Autónoma de
Madrid).
9.
REFERENCES
Friesen, N., 2005, Interoperability and learning objects: An overview of e-learning
standardization. Interdisciplinary J. of Knowledge and Learning Objects. 1: 23-31.
Lama, N., Arroyo, S., Sicilia, M.A. and López Cobo, J.M., 2006, Making learning objects
useful with semantic web services. eChallenges Conf.
Lytras, M. and Naeve, A., (eds), 2005, Intelligent Learning Infrastructures for Knowledge
Intensive Organizations: A Semantic Web Perspective, IDEA publishing group.
McGreal, R., 2004, Learning objects: A Practical definition. Int. J. of Instructional
Technology and Distance Learning 1(9).
38
http://www.win.tue.nl/SW-EL/
Monceaux, A., Guss, J., 2006, Training management system for aircraft engineering: indexing
and retrieval of corporate learning object. Industry Forum: Business applications of
Semantic Web challenge Research in the European Semantic Web Conference ESWC’06.
Naeve, A. and Sicilia, M-A., 2006, Learning processes and processing learning: From
Organizational Needs to Learning Designs, Invited paper for the ADALE workshop (in
conjunction with the Adaptive Hypermedia conference).
Naeve, A., Yli-Luoma, P., Kravcik, M., Lytras, M., Simon, B., Lindegren, M., Nilsson, M.,
Palmér, M., Korfiatis, N., Wild, F., Wessblad, R., Kamtsiou, V., Pappa, D. and Kieslinger,
B., 2005, A Conceptual Modelling Approach to Studying the Learning Process with a
Special Focus on Knowledge Creation. Deliverable 5.3 of the Prolearn EU/FP6 Network
of Excellence, IST 507310.
Rothwell, W. and Kazanas, H., 1992, Mastering the Instructional Design Process. JosseyBass.
Sicilia, M. A., 2005, Ontology-based competency management: Infrastructures for the
knowledge- intensive learning organization, in Intelligent Learning Infrastructures in
Knowledge Intensive Organizations: A Semantic Web Perspective. Idea Group, Hershey,
PA, pp. 302-324.
Sicilia, M.A. and García-Barriocanal, E., 2005, On the convergence of formal ontologies and
standardized e-learning. Journal of Distance Education Technologies, 3(2): 12-28
Sicilia, M.A. and Lytras, M., 2005, The semantic learning organization. The learning
organization, 12(5): 402-410.
Sicilia, M.A. (ed.), 2006. Competencies in organizational e-learning. Concepts and Tools.
Idea Group publishing, Hershey, PA.