<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.0 20120330//EN" "JATS-archivearticle1.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink">
  <front>
    <journal-meta />
    <article-meta>
      <title-group>
        <article-title>The Standards Landscape for AI-based Guided Experiential Learning</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <string-name>Robby Robson</string-name>
          <xref ref-type="aff" rid="aff0">0</xref>
        </contrib>
        <aff id="aff0">
          <label>0</label>
          <institution>Eduworks Corporation</institution>
          ,
          <addr-line>400 SW 4th Street, STE 110, Corvallis, OR</addr-line>
          ,
          <country country="US">USA</country>
        </aff>
      </contrib-group>
      <fpage>25</fpage>
      <lpage>32</lpage>
      <abstract>
        <p>Guided Experiential Learning (GEL) [1] is a pedagogical framework in which proficiency is gained through focused, repetitive practice under real world or simulated real world conditions. It is increasingly implemented with the aid of games, simulations, virtual, augmented, and mixed reality. In these implementations, data from learning environments is used in algorithms and AI models that evaluate performance, estimate learner states, and support instructors or adaptive instructional systems in selecting scenarios that provide learners with targeted practice under a targeted set of conditions. Underlying these implementations are multi-stage data strategies that rely heavily on data and data exchange standards, as is illustrated by the Synthetic Training Environment Experiential Learning - Readiness (STEEL-R) project [2, 3]. Relevant standards include standards published by IEEE, 1EdTech, W3C, and other standard development organizations as well as standards associated with the US Advanced Distributed Learning (ADL) Initiative's Total Learning Architecture [4, 5]. This paper outlines the STEELR data strategy as an example of a GEL implementation, categorizes relevant standards, and discusses how they are used in STEEL-R, and suggests possible future enhancements that will be needed both for AI-enabled GEL systems.</p>
      </abstract>
      <kwd-group>
        <kwd>eol&gt;Guided Experiential Learning</kwd>
        <kwd>Standards</kwd>
        <kwd>Learning Technology</kwd>
        <kwd>AI</kwd>
        <kwd>Experience Orchestration</kwd>
        <kwd>Competencies</kwd>
        <kwd>Learner Models</kwd>
        <kwd>Adaptive Instructional Systems</kwd>
        <kwd>Synthetic Training Environment 1</kwd>
      </kwd-group>
    </article-meta>
  </front>
  <body>
    <sec id="sec-1">
      <title>1. A Motivating Example</title>
      <p>
        We start our discussion of standards for AI-based GEL by giving an overview of the Synthetic
Training Environment Experiential Learning – Readiness (STEEL-R) [
        <xref ref-type="bibr" rid="ref2 ref3">2, 3</xref>
        ]. STEEL-R was developed
to support U.S. Army training using a combination of syn-thetic (i.e., game-based), semi-synthetic (i.e.,
mixed reality), and live training environments. It takes a competency-based approach in which (1)
performance is evaluated on tasks, activities, and behaviors in training scenarios, (2) evaluations
generate assertions [
        <xref ref-type="bibr" rid="ref6">6</xref>
        ] about competencies that learners have demonstrated, (3) assertions are used to
generate competency profiles that estimate the level of competency attained with respect to
competencies in a competency framework, and (4) competency levels are tracked over time and used
to inform the selection of training. Fig. 1 shows this process from a functional perspective.
      </p>
    </sec>
    <sec id="sec-2">
      <title>STEEL-R Data Strategy</title>
      <p>
        Underlying the functional perspective in Fig. 1 is a data strategy. In this strategy, the Generalized
Intelligent Framework for Tutoring (GIFT) [
        <xref ref-type="bibr" rid="ref7">7</xref>
        ] connects to training systems via gateway modules. GIFT
observes events in these systems and evaluates performance using algorithms programmed into its
Domain Knowledge File [
        <xref ref-type="bibr" rid="ref8">8</xref>
        ]. GIFT emits experience API (xAPI) statements [
        <xref ref-type="bibr" rid="ref9">9</xref>
        ] that encode these
evaluations. These statements are stored and filtered in Learning Record Stores, from where they are
retrieved by the Competency and Skills System (CaSS) [
        <xref ref-type="bibr" rid="ref10 ref11 ref12">10–12</xref>
        ]. A decoder in CaSS translates them
into assertions that a competency estimator uses to compute longitudinal competency profiles. These
profiles estimate and track competency and skills progression over time and, together with a catalog of
available scenarios from an experience index, are used by a Navigator that identifies potential training
experiences based on user criteria. Included in STEEL-R is an experience design tool that produces
experience training support packages (XTSPs). XTSPs define training experiences in a format designed
to be ingested by training systems. This data strategy is shown in Fig. 2.
1.2.
      </p>
    </sec>
    <sec id="sec-3">
      <title>AI in STEEL-R</title>
      <p>As of the writing of this paper, STEEL-R uses deterministic methods for converting performance
evaluations to assertions, using assertions to estimate competencies, and recommending training
experiences. All of these functions, however, are designed with AI in mind and are anticipated to
involve AI and ML in the next iteration.
1.3.</p>
    </sec>
    <sec id="sec-4">
      <title>Relevant Standards</title>
      <p>
        Turning to the main topic of this paper, multiple standards represent and communicate data in
STEEL-R (see Fig. 2). These include standards used in the Advanced Distributed Learning (ADL)
initiative’s Total Learning Architecture (TLA) [
        <xref ref-type="bibr" rid="ref5">5</xref>
        ], data formats in CaSS, and standards for competency
definitions and competency frameworks [
        <xref ref-type="bibr" rid="ref13 ref14">13, 14</xref>
        ]. Most of these standards may be classified as learning
technology standards, which we discuss next.
      </p>
    </sec>
    <sec id="sec-5">
      <title>2. Learning Technology Standards</title>
      <p>
        Since circa 1997, many standards development organizations (SDOs) have produced standards
intended to support the development, deployment, and operation of learning technologies. Leading such
SDOs include the Aviation Industry Computer-Based Training Committee (AICC, closed after 25+
years of operation [
        <xref ref-type="bibr" rid="ref15">15</xref>
        ]), the IMS Global Learning consortium (now 1EdTech) [
        <xref ref-type="bibr" rid="ref16">16</xref>
        ], the IEEE Learning
Technology Standards Committee [
        <xref ref-type="bibr" rid="ref17">17</xref>
        ], ISO/IEC JTC1 SC36 (information technology for learning,
education and training) [
        <xref ref-type="bibr" rid="ref18">18</xref>
        ], and the European Committee for Standardization (CEN) Workshop on
Learning Technology (CEN-WSLT) which also is no longer operational. As of 2014, an observatory
maintained by CEN-WSLT listed over 50 learning technology standards in areas ranging from
accessibility and assessment to runtime and vocabulary [19], and many more have been developed since
then.
      </p>
      <p>Standards that support general learning technologies apply to GEL systems, but in many cases
require modifications, extensions, or new components. This is because most existing learning
technology standards evolved from formal education and regulatory training. The standards that
prevailed were, as is often the case, the ones that were most market-relevant, and in this case the market
involved Learning Management Systems (LMSs) that delivered education and training in cognitive
domains and assessed learners at a single point in time. GEL, in contrast, tends to require longitudinal
assessment, multi-modal delivery, and non-cognitive competencies and skills. Richer data must be
tracked, computational models may include components such as training conditions and records of
practice, and multi-dimensional competencies and skills frameworks are in-volved. Moreover, as
Artificial Intelligence (AI) and Machine Learning (ML) become more central to the operation of GEL
systems, the properties of learning experiences that matter will shift from those that help human
operators catalog them to those needed by AI-driven recommendation and sequencing engines. As we
go through the list of most relevant learning technology standards, we will point out where and how
changes should or may be made to support GEL.
2.1.</p>
    </sec>
    <sec id="sec-6">
      <title>Reporting (also known as Tracking) Standards</title>
      <p>The most prevalent standards that track and report on student activities in learning systems are AICC
Computer Managed Instruction (CMI) standards [20], the Shareable Object Reference Model
(SCORM) [21], xAPI, and IMS Caliper Analytics (Caliper) [22]. The first two – AICC and SCORM –
were LMS-centric standards designed to report what a student completed and the results of quizzes and
tests. xAPI and Caliper replace these with standards that can be used for other types of reporting, but
when deployed as replacements for AICC or SCORM, they are usually configured to report the same
data, i.e., completions of exercises and learning units and the results of formative and summative
assessments on cognitive tasks.</p>
      <p>GEL requires systems to report performance on complex tasks and behaviors, not just formative and
summative assessments. As explained in [23–25], xAPI can do this with the aid of properly designed
xAPI profiles [26] that enable more varied sets of verbs and contexts to appear in xAPI statements, and
Caliper can be extended in similar ways. Developing an appropriate xAPI profile was a key enabler of
the STEEL-R data strategy, and it is likely that much of the GEL standardization efforts around tracking
and reporting will focus on the development of profiles which themselves may be viewed as standards
for a particular type of GEL system or application domain. If this comes to fruition, then registries of
profiles will be needed. Such registries have been set up commercially [27] and by the ADL [28] but to
the best of the author’s knowledge have not become commonplace or standardized.
2.2.</p>
    </sec>
    <sec id="sec-7">
      <title>Experience Orchestration</title>
      <p>When AICC and SCORM were developed, learning experiences were selected, sequenced, and
delivered by an LMS. Selection and sequencing could be governed by instructions contained in AICC
and SCORM packages (see Section 2.3). In the world of AI-enabled intelligent tutoring systems (ITS)
and, more generally, adaptive instructional systems (AIS) [29], progressions through topics and learning
experiences are governed by algorithms that implement the outer loop [30], for which we know of no
widely adopted standard. Nonetheless, the outer loop generally applies to a single ITS or AIS, whereas
in STEEL-R and the training environments it is meant to support, relatively granular learning
experiences are delivered by multiple systems, and outer loop style adaptation is accomplished by
GIFT, which also can configure training as it is in progress. We propose that this process be called
experience orchestration, or XO.
xAPI and Caliper have moved beyond an LMS-centric view of learning technology by supporting
decentralized networks of learning systems. They do not, however, ad-dress how experiences are
selected or sequenced. In STEEL-R, XO instructions given to GIFT come from human operators. We
believe this will soon be replaced enhanced or replaced by AI-driven XO. For this purpose, two
capabilities will be necessary:
1. An interoperable means of expressing XO patterns or instructions so that systems such as GIFT
can understand and execute them; and
2. A standardized way of expressing the properties of learning experiences, learning goals, and
learners that the AI examines when making XO decisions.</p>
      <p>With regard to the first capability, many researchers have proposed methods of representing learning
paths [31–35], and a schema for expressing learning pathways was developed by the Credential Engine
as part of its Credential Transparency Description Language (CTDL) [36]. These proposals, as well as
the CTDL schema, operate at the level of courses and credentials and not at the more granular level of
learning experiences. At the other end of the granularity spectrum, ITS sequence knowledge
components (KCs) based on programmable instructions [37, 38], although as mentioned earlier not in
any standardized manner. There seems to be a need for standards for XO instructions that operate at the
level of learning experiences (or “micro-learning”) that are finer grained than courses and larger than
the instruction associated with single KCs. In passing, we remark that this requirement is reminiscent
of the attempts made circa 2010 to use Business Process Execution Language (BPEL) to instantiate
IMS Learning Design [39–41] and that, to the best of our knowledge, did not enjoy real-world
commercial adoption. In our view, a standards are still needed to represent experience orchestration
rules that depend on the properties of learning experiences, learning goals, and learner models – the
latter of which are discussed in Section 2.4.</p>
      <p>With regard to the properties of learning experiences, standards for learning object metadata were
developed by the IEEE, IMS Global, ISO/IEC JTC1 SC36, the Dublin Core Metadata Initiative [42],
and others precisely for the purpose of identifying and communicating these properties in interoperable
ways. The problem faced in applying these to GEL is identifying which properties should be expressed
more than it is developing new ways to express them. Moreover, there is ongoing activity in this area.
For example, the IEEE standard for learning object metadata is in the process of being up-dated [43]
and was used in STEEL-R with extensions that allow properties such as training conditions, available
stressors, and difficulty factors to be expressed. CTDL, mentioned earlier, is naturally concerned with
the properties of credentials but includes an extensive set of properties of learning experiences from
which credentials are obtained. Schema such as the Creative Work schema hosted by Schema.org [44]
are easily extended to include properties relevant to learning applications, as was done by the Learning
Resource Metadata Initiative (LRMI) [45].
2.3.</p>
    </sec>
    <sec id="sec-8">
      <title>Packaging</title>
      <p>Packaging standards such as 1EdTech Common Cartridge [46] and SCORM manifests define
content aggregations that can be loaded into an LMS or a learning environment. They identify the
content to be delivered and include additional information, such as metadata describing the content and
sequencing instructions. They are the analog of the recorded media that existed at the time those
standards were developed and that could be inserted into a player and played, with a more modern
analog being software containers [47].</p>
      <p>For GEL applications, packages should include experience orchestration (see Section 2.2), and
information about what will be practiced, how often it will be practiced, under what conditions it will
be practiced, and how performance will be evaluated. For STEEL-R, researchers at the University of
Texas at Austin developed an XTSP data format that represents experiences in synthetic learning
environments. Earlier attempts at such representations, including IMS Learning Design [48], did not
succeed in creating packages that could be delivered by multiple learning systems, but XTSP has
promise. We expect that standardized abstract representations of learning experiences along the lines
of XTSP will play a crucial role in GEL.
2.4.</p>
    </sec>
    <sec id="sec-9">
      <title>Interoperable Learner Models</title>
      <p>A standardized interoperable form of a learner model [30] has long been a goal of the ITS/AIS
community. This would allow any conformant AIS to read and update the data it uses to adapt learning
to the needs of an individual and to export these data for use by the next AIS. The question this raises
is what data are these?</p>
      <p>A partial answer to this question is that the data needed to adapt learning to an individual’s needs is
the learner’s competency profile, i.e., the list of competencies, skills, capabilities, traits, etc. possessed
by a learner together with the level at which each one is possessed. For such a profile to be
machineactionable, it must point to machine-actionable representations of competencies and competency
frameworks, which is what CaSS provides as linked data in STEEL-R and in other implementations.</p>
      <p>For GEL, snapshots of the competencies held by a learner are not sufficient. Since GEL requires
that systems identify and deliver deliberate episodic practice at optimal intervals, learner models must
have a time dimension, and quantities such as past practice should be included. The STEEL-R version
of CaSS outputs competency profiles of this type and can associate values in user-defined concept
schema to competencies. This presents a standardization opportunity that we believe would be of
significant benefit to all AIS.
2.5.</p>
    </sec>
    <sec id="sec-10">
      <title>Assertions and Digital Credentials</title>
      <p>A core notion in STEEL-R is that of an assertion. In STEEL-R these are stored and processed in
CaSS, and to the best of our knowledge, the abstraction of an assertion in this form first appeared in
CaSS. The term is now also used by 1EdTech in its Comprehensive Learner Record (CLR) specification
[49], although in CLR assertions are about achievements, and achievements include accomplishments
such as the completion of a degree or course as well as evidence of a competency. CaSS assertions are
strictly about the possession or demonstration of a competency in an identified competency framework.
Assertions are expressed in a standardized format within CaSS, but this format has not been
standardized by any SDO.</p>
      <p>An area where SDOs are actively involved is digital credentials and electronic learner records.
Relevant standards include W3C verifiable credentials [50], Open Badges [51], Comprehensive Learner
Records [49], and Learner and Employment Records [52]. These provide historical records that can be
converted into competency assertions and combined with other evidence when estimating competency
states. For use in GEL, credentials and records should include accurate timestamps and identify the
type, conditions, and frequency of relevant practice, as is done for pilot licenses. This is necessary to
create assertions of the type used in STEEL-R is not currently the case. It is not clear whether GEL will
be considered as credentialing and badging standards mature, but we hope that it will be.</p>
    </sec>
    <sec id="sec-11">
      <title>3. Privacy, Ethics, and Security</title>
      <p>Finally, there are many non-learning technology standards that are likely to affect the design of GEL
systems, especially if they incorporate AI. For example, the IEEE Standards Association has released
several standards as part of its global initiative on the ethics of autonomous and intelligent systems [53],
and is developing a related certification program [54]. It has also published a standard for an
ageappropriate digital services framework [55], and around the world governments are passing legislation
that affects data privacy rights [56] and the use of AI [57]. Standards and regulations of this nature are
likely to proliferate in response to concerns about generative AI. In the GEL context, they are relevant
to the design, development, and deployment of GEL systems and, regardless of their appropriateness,
could present challenges to the ability of such systems to collect and process data.</p>
      <p>Security is another area where regulations could affect the future of GEL. To quote from the EU
Cyber Resilience Act (CRA) website [59], “From baby-monitors to smart-watches, products and
software that contain a digital component are omnipresent in our daily lives. Less apparent to many
users is the security risk such products and soft-ware may present.” The instrumentation used to collect
data for GEL, including sensors and software (e.g., for virtual or mixed reality) will fall under the CRA,
and in military and corporate settings, the systems used for GEL may be required to conform to
standards such as the NIST 800-series standards [60].</p>
      <p>Ignoring regulatory environments and concerns about issues such as privacy, AI ethics, and security
is not a recipe for success, as was poignantly illustrated in the educational technology community by
the collapse of In Bloom in 2014 [58]. We believe it is critical that the GEL community (and the
educational and training technology in general) work with the government, non-governmental, and
standards development organizations that are creating the regulatory and standards environments. GEL
requires examining a leaner’s history, and GEL systems must collect more extensive data than most
existing learning systems. This may make GEL systems more sensitive to privacy, ethical, and security
concerns than more traditional learning systems. It is therefore important that the GEL community have
a voice in the development of privacy, ethics and security standards and that they incorporate them into
their own standards.</p>
    </sec>
    <sec id="sec-12">
      <title>4. References</title>
      <p>[19] CEN-WSLT (2014) LTSO - Learning Technology Standards Observatory :: In: CEN Workshop
on Learning Technology.
0https://web.archive.org/web/20141114201940/http://www.cenltso.net/main.aspx?AspxAutoDetectCookieSupport=1. Accessed 2 Jul 2023
[20] AICC (1998) A980519C.pdf
[21] ADL Sharable Content Object Reference Model (SCORM®). In: ADL Initiative.</p>
      <p>https://adlnet.gov/past-projects/scorm/. Accessed 2 Jul 2023
[22] 1EdTech Caliper Analytics. https://www.imsglobal.org/activity/caliper. Accessed 2 Jul 2023
[23] Blake-Plock S, Hoyt W, Casey C (2021) Instrumenting GIFT with xAPI: a use case for IEEE
P9274.3.x standards activity and implications for the broader field of ITS and AIS. In: Goldberg
B (ed) Proceedings of the 9th Annual GIFT Users Symposium (GIFTSym9). U.S. Army Combat
Capabilities Development Command – Soldier Center., pp 55–59
[24] Robson R, Ray F, Hernandez M, et al (2022) Mining Artificially Generated Data to Estimate</p>
      <p>Competency
[25] Blake-Plock S, Hoyt W, Casey C, Zapata-Rivera D (2020) Data Analytics and Visualization for
xAPI Learning Data: Considerations for a GIFT Strategy. Design Recommendations for Intelligent
Tutoring Systems 163–171
[26] ADL (2018) xAPI-profiles: A set of documents addressing the structure of and supporting services
for xAPI Profiles. https://github.com/adlnet/xapi-profiles. Accessed 27 Feb 2022
[27] Rustici Software (2023) XAPI Registry. In: xapi.com. https://xapi.com/registry/. Accessed 3 Jul
2023
[28] ADL (2023) ADL xAPI Profile Server. https://profiles.adlnet.gov/. Accessed 3 Jul 2023
[29] Sottilare RA, Schwarz J (2020) Adaptive Instructional Systems. Springer International Publishing
[30] VanLehn (2006) The behavior of tutoring systems. Int J Artif Intell Educ
[31] Janssen J, Berlanga A, Vogten H, Koper R (2008) Towards a learning path specification.</p>
      <p>International Journal of Continuing Engineering Education and Life Long Learning 18:77–97
[32] Ramos DB, Ramos IMM, Gasparini I, de Oliveira EHT (2021) A New Learning Path Model for</p>
      <p>E-Learning Systems. IJDET 19:34–54
[33] Sanchez Nigenda R, Maya Padrón C, Martínez-Salazar I, Torres-Guerrero F. (2018) Design and
evaluation of planning and mathematical models for generating learning paths. Comput Intell
34:821–838
[34] Salinas-Ibáñez J, De-Benito B (2020) Construction of personalized learning pathways through
mixed methods. Comunicar 28:31–42
[35] Concentric Sky (2018) Open pathways for learning. In: 1EdTech.</p>
      <p>http://www.imsglobal.org/concentric-sky-open-pathways. Accessed 3 Jul 2023
[36] Credential Engine (2022) New CTDL pathway constraints terms June release. In: Credential
Engine.
https://credentialengine.org/2022/07/06/new-ctdl-pathway-constraints-terms-junerelease/. Accessed 3 Jul 2023
[37] Corbett AT, Anderson JR (1994) Knowledge tracing: Modeling the acquisition of procedural
knowledge. User Model User-adapt Interact 4:253–278
[38] Aleven V, McLaren BM, Sewall J, Koedinger KR (2009) A new paradigm for intelligent tutoring
systems: Example-tracing tutors. Int J Artif Intell Educ 19:105–154
[39] Chen C-T, Cheng YC, Hsu T-S, et al (2009) Delivering specification-based learning processes
with service-oriented architecture: A process translation approach.
https://citeseerx.ist.psu.edu/document?repid=rep1&amp;type=pdf&amp;doi=6f2606e6baf9
1d28497be73a7f53108f8b5ac93e. Accessed 4 Jul 2023
[40] Sandu V, Nedelcu A-V, Cocorada SA, Dinu S (2010) Instructional design in the vocational training
on “computer networking.” In: 2010 12th International Conference on Optimization of Electrical
and Electronic Equipment. ieeexplore.ieee.org, pp 1084–1090
[41] Kanagwa B, Agaba J, Tuheirwe D, Lunkuse S (2011) Improving Learning objects reusability
through automatic generation web services
[42] DCMI (2023) Dublin Core Metadata Initiative. https://www.dublincore.org/. Accessed 3 Jul 2023
[43] IEEE (2020) Standard for Learning Object Metadata
[44] Schema.org CreativeWork - schema.org Type. https://schema.org/CreativeWork. Accessed 30
Mar 2020
[45] DCMI (2023) About LRMI. In: Dublin Core Metadata Initiative.</p>
      <p>https://www.dublincore.org/about/lrmi/. Accessed 4 Jul 2023
[46] 1EdTech (2023) Common Cartridge. https://www.imsglobal.org/activity/common-cartridge.</p>
      <p>Accessed 2 Jul 2023
[47] Wikipedia (2023) Containerization (computing). In: Wikipedia, The Free Encyclopedia.</p>
      <p>https://en.wikipedia.org/w/index.php?title=Containerization_(computing)&amp;oldid =1163448757
[48] Olivier B, Tattersall C (2005) The Learning Design Specification. In: Koper R, Tattersall C (eds)
Learning Design: A Handbook on Modelling and Delivering Networked Education and Training.</p>
      <p>Springer Berlin Heidelberg, Berlin, Heidelberg, pp 21–40
[49] Comprehensive Learner Record | IMS Global Learning Consortium.</p>
      <p>https://www.imsglobal.org/activity/comprehensive-learner-record. Accessed 26 Apr 2020
[50] Sporny M, Longley D, Chadwick D (2022) Verifiable credentials data model v1.1. In: W3C</p>
      <p>Verifiable Credentials. https://www.w3.org/TR/vc-data-model/. Accessed 13 Jun 2023
[51] 1EdTech (2023) Open Badges Home Page. In: Open Badges. https://openbadges.org/. Accessed
13 Jun 2023
[52] America’s R Learning and employment records, a pathway for.
https://americafirstpolicy.com/assets/uploads/files/learning-and-employment-records-a-pathwayfor-restoring-americas-workforce.pdf. Accessed 12 Jan 2023</p>
    </sec>
  </body>
  <back>
    <ref-list>
      <ref id="ref1">
        <mixed-citation>
          [1]
          <string-name>
            <surname>Goldberg</surname>
            <given-names>B</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Robson</surname>
            <given-names>R</given-names>
          </string-name>
          (
          <year>2023</year>
          )
          <article-title>AI to Support Guided Experiential Learning</article-title>
          . In: to appear
          <source>in Proceedings of AI in Education 2023</source>
          . Springer
        </mixed-citation>
      </ref>
      <ref id="ref2">
        <mixed-citation>
          [2]
          <string-name>
            <surname>Goldberg</surname>
            <given-names>B</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Owens</surname>
            <given-names>K</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Gupton</surname>
            <given-names>K</given-names>
          </string-name>
          , et al (
          <year>2021</year>
          )
          <article-title>Forging Competency and Proficiency through the Synthetic Training Environment with an Experiential Learning for Readiness Strategy</article-title>
          .
          <source>National Training and Simulation Association</source>
        </mixed-citation>
      </ref>
      <ref id="ref3">
        <mixed-citation>
          [3]
          <string-name>
            <surname>Hernandez</surname>
            <given-names>M</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Blake-Plock</surname>
            <given-names>S</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Owens</surname>
            <given-names>K</given-names>
          </string-name>
          , et al (
          <year>2022</year>
          <article-title>) Enhancing the Total Learning Architecture for Experiential Learning</article-title>
          .
          <source>National Training and Simulation Association</source>
        </mixed-citation>
      </ref>
      <ref id="ref4">
        <mixed-citation>
          [4]
          <string-name>
            <surname>ADL</surname>
          </string-name>
          (
          <year>2022</year>
          )
          <article-title>Understanding the TLA reference implementation</article-title>
          . In:
          <article-title>Advanced Distributed Learning Initiative</article-title>
          . https://adlnet.gov/guides/tla/service-definitions/TLA-ReferenceImplementation.
          <source>html. Accessed 28 Feb 2022</source>
        </mixed-citation>
      </ref>
      <ref id="ref5">
        <mixed-citation>
          [5]
          <string-name>
            <surname>Gordon</surname>
          </string-name>
          , Hayden, Johnson, Smith (
          <year>2020</year>
          )
          <article-title>Total Learning Architecture 2019 Report</article-title>
          . Advanced Distributed Learning
        </mixed-citation>
      </ref>
      <ref id="ref6">
        <mixed-citation>
          [6]
          <string-name>
            <surname>Robson</surname>
            <given-names>R</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Poltrack</surname>
            <given-names>J</given-names>
          </string-name>
          (
          <year>2017</year>
          )
          <article-title>Using competencies to map performance across multiple activities</article-title>
          .
          <source>In: Proceedings of the I/ITSEC</source>
          . adlnet.gov
        </mixed-citation>
      </ref>
      <ref id="ref7">
        <mixed-citation>
          [7]
          <string-name>
            <given-names>U.S.</given-names>
            <surname>Army DEVCOM Soldier Center</surname>
          </string-name>
          (
          <year>2022</year>
          )
          <article-title>GIFT Portal</article-title>
          . https://www.gifttutoring.org/projects/gift/wiki/Overview. Accessed 8 Jun 2022
        </mixed-citation>
      </ref>
      <ref id="ref8">
        <mixed-citation>
          [8]
          <string-name>
            <surname>GIFT</surname>
          </string-name>
          (
          <year>2021</year>
          )
          <article-title>Domain knowledge file 2021-2</article-title>
          . In: GIFT Tutoring Portal. https://www.gifttutoring.org/projects/gift/wiki/Domain_Knowledge_File_
          <fpage>2021</fpage>
          -
          <lpage>2</lpage>
          . Accessed 27 Feb 2022
        </mixed-citation>
      </ref>
      <ref id="ref9">
        <mixed-citation>
          [9]
          <string-name>
            <surname>ADL</surname>
          </string-name>
          (
          <year>2021</year>
          )
          <article-title>Experience API (xAPI) standard</article-title>
          . In: ADL Initiative. https://adlnet.gov/projects/xapi/.
          <source>Accessed 6 Jun 2022</source>
        </mixed-citation>
      </ref>
      <ref id="ref10">
        <mixed-citation>
          [10]
          <string-name>
            <surname>Eduworks</surname>
          </string-name>
          (
          <year>2020</year>
          )
          <article-title>The Competency and Skills System</article-title>
          . In: CaSS. http://cassproject.org. Accessed 18 May 2020
        </mixed-citation>
      </ref>
      <ref id="ref11">
        <mixed-citation>
          [11]
          <string-name>
            <surname>Havas</surname>
            <given-names>K</given-names>
          </string-name>
          (
          <year>2021</year>
          )
          <article-title>CaSS Authoring Tools</article-title>
          . DTIC
        </mixed-citation>
      </ref>
      <ref id="ref12">
        <mixed-citation>
          [12]
          <string-name>
            <surname>ADL</surname>
          </string-name>
          (
          <year>2020</year>
          )
          <article-title>Competency &amp; Skills System (CaSS)</article-title>
          . In:
          <article-title>Advanced Distributed Learning Initiative</article-title>
          . https://adlnet.gov/projects/cass/.
          <source>Accessed 4 Apr 2020</source>
        </mixed-citation>
      </ref>
      <ref id="ref13">
        <mixed-citation>
          [13]
          <string-name>
            <surname>IEEE</surname>
          </string-name>
          (
          <year>2022</year>
          )
          <article-title>IEEE Approved Draft Standard Data Model for Sharable Competency Definitions</article-title>
          .
          <source>IEEE Standards Association</source>
          , 501 Hoes Lane, Piscataway, NJ 08854
        </mixed-citation>
      </ref>
      <ref id="ref14">
        <mixed-citation>
          [14]
          <string-name>
            <surname>IEEE</surname>
          </string-name>
          (
          <year>2022</year>
          )
          <article-title>IEEE Recommended Practice for Defining Competencies</article-title>
          .
          <source>IEEE Standards Association</source>
          , 501 Hoes Lane, Piscataway, NJ 08854
        </mixed-citation>
      </ref>
      <ref id="ref15">
        <mixed-citation>
          [15]
          <string-name>
            <surname>AICC</surname>
          </string-name>
          (
          <year>2023</year>
          )
          <article-title>Our Journey is at an End. In: Aviation Industry Computer-Based Training Committee</article-title>
          . https://www.aicc.org/.
          <source>Accessed 2 Jul 2023</source>
        </mixed-citation>
      </ref>
      <ref id="ref16">
        <mixed-citation>
          [16] (
          <year>2023</year>
          )
          <article-title>1EdTech</article-title>
          . https://www.1edtech.org/.
          <source>Accessed 2 Jul 2023</source>
        </mixed-citation>
      </ref>
      <ref id="ref17">
        <mixed-citation>
          [17]
          <string-name>
            <surname>IEEE</surname>
            <given-names>SA</given-names>
          </string-name>
          (
          <year>2023</year>
          )
          <article-title>IEEE Learning Technology Standards Committee (LTSC) -Enabling better learning through global technical standards</article-title>
          . In: IEEE LTSC. https://sagroups.ieee.org/ltsc/.
          <source>Accessed 2 Jul 2023</source>
        </mixed-citation>
      </ref>
      <ref id="ref18">
        <mixed-citation>
          [18]
          <string-name>
            <surname>ISO</surname>
          </string-name>
          (
          <year>2023</year>
          )
          <article-title>Information technology for learning, education and training</article-title>
          .
          <source>In: ISO/IEC JTC1 SC36</source>
          . https://www.iso.org/committee/45392.html.
          <source>Accessed 2 Jul 2023</source>
        </mixed-citation>
      </ref>
    </ref-list>
  </back>
</article>