CHUNK:Curated Heuristic Using a Network of Knowledge

Ralucca Gera

Department of Applied Mathematics,Associate Provost for Graduate Education,

Teaching and Learning Commons,Naval Postgraduate School, Monterey, CA

Email: rgera@nps.edu

Michelle L. Isenhour

Department of Operations Research,Naval Postgraduate School,

Monterey, CAEmail: mlisenho@nps.edu

D’Marie Bartolf

Teaching and Learning Commons,Naval Postgraduate School,

Monterey, CAEmail: dmarie.bartolf.ctr@nps.edu

Simona Tick

Graduate School of Business and Public Policy,Naval Postgraduate School,

Monterey, CAEmail: sltick@nps.edu

Abstract—What is the potential of a 21st century learningenvironment that mirrors personalized Apps? In contrast tothe standard linear or tree-like educational system of sequentiallectures or chapters, we propose a real-time, modular, adaptiveteaching-learning environment for enhanced and personalizededucation, called the Curated Heuristic Using a Network ofKnowledge (CHUNK) Learning. This CHUNK Learning modelbreaks away from the predictable pattern of traditional edu-cation models and provides content delivery that adapts to thecapabilities, learning styles, and approaches to problem-solvingof every learner. With CHUNK Learning, students are em-powered by a student-centered teaching-learning system whosepurpose is to make learning engaged, flexible, and respectful ofthe students’ time. Much like a computer game approach, theCHUNK Learning maintains an on-line learner profile for eachuser, which guides the learner through a Network of Knowledgecomposed of lesson materials joined together by prerequisitesand common attributes based on competency or skill levels.Our vision also includes a mix of different delivery methods,to include demos, videos, interactive applications, TED talks,webinars, programming languages (personalized to the skill ofthe learner), and data manipulations, all delivered through acombination of online and traditional experiences.

I. INTRODUCTION

Traditional education places all students through the sametopics, at the same time, at the same pace, generally teachingto the average student. However, such a system fails toprovide every student the appropriate opportunity to learnand engage, leaving some learners struggling, while failingto challenge others. A strength in student-centered learning,as opposed to traditional education, is the opportunity forindependence and autonomy that enhances different skills,abilities, and interests learners might have. Unlike most

established sciences, new research fields such as NetworkScience (an emerging field overseeing traditional computerscience, operations research, sociology, and other fields)hint at non-traditional methods of education that serve asa starting point for the current effort.

The contrast between our views on the potential of 21stcentury education and the current educational status quogenerates ideas which we believe can improve effectivenessof educational programs through enhanced retention, trans-fer ability and critical thinking linked to learner-centerededucation. We introduce a vision for an adaptive teaching,flexible learning, and technology-enhanced student-centerededucation strategy for the 21st century learner that operatesin the big data environment [2, 3, 28]. We describe our viewon how different teaching and learning tools can be joinedinto a web/network of knowledge that will individualizelearning and leave a persistent impact on the learner’s career.We introduce our unique structure and vision for catering tothe needs, motivations, and supply of learners with a systemthat:

• Offers intense, short, and focused educational modules• Stimulates interest and relevance of topics• Integrates new information with learner’s pre-existing

knowledge• Provides personalized and individualized education• Optimizes content and methodology delivery to meet

the needs of each learner

The structure of the paper is as follows. We present relatedwork in Section II followed by a short overview vision

of what the system is becoming in Section III. We thenintroduces the CHUNK framework in Section IV followedby the science behind CHUNK Learning in Section V. Wepresent the Network of Knowledge and its components inSections VI and VII, respectively. Furthermore, we discussand run through an example of how CHUNK Learning isindividualized and personalized in Sections VIII and IX. Weconclude with further direction in Section X.

II. RELATED WORK

In the early 90’s, education slowly started to incorporatedigital resources and use of Internet in the classrooms.By 1999, 80 percent of the schools used these resourcesin some form [13]. In the same year, studies of the In-ternet surfaced and network science research started tobloom [5, 12]. Yet, to this day, we still use the Internet anddigital resources primarily as a resource for information orto host distance learning and Massive Open Online Courses(MOOCs). Below, we expand on the richer use of pullingeducational information to personalize educational contentand to position the learner the way Amazon, Netflix andsocial media do.

Educational content is generally organized in a hierarchi-cal structure following a textbook or a collection of books.Students enroll in a curriculum that consists of severalcourses. Each course has themes organized into chapters,and each chapter consists of several sections. In general, thesmallest level of educational material relevant for learningis a section, whereas the smallest level a student can choosefrom is at the course level. Our vision is to consider thedata at a more granular level at the tier of topics, wherea section could be comprised of one or several topics,possibly captured on transcripts by micro-credits. More-over, based on design cognition, visualizing the structuredrepresentation of concepts and the relations between themas a map allows the learner to generate a deeper, longerlasting learning [21]. Concept mapping allows students tolearn in a more meaningful way by elevating their previousexperiences and knowledge and relating them to the newconcepts for a more complex, longer lasting, more effectinglearning [16].

Information theory shows that learners are limited bythe amount of information one can receive, process, andremember at the time [20]. Techniques such as “several stim-ulus dimensions, recording, and various mnemonic devices”along with breaking down the information supports thelearning environment. One of these techniques is to presentthe information in smaller and shorter and interconnectedchunks that has been already used for both traditional [17]and for e-learning [32].

A step forward from traditional education is web-basededucation or e-learning [4, 18, 24, 26]. Particularly, e-learning received more attention in the form of MassiveOpen Online Courses (MOOCs) [22, 7]. While the digiti-zation of the lectures gives students freedom in interactingwith the content whenever they want, all the students are

still exposed to the exact same information, presented in asingle way, at a single pace.

Recent studies indicate that e-learning efficiency can bedramatically improved if personal prerequisites, skills, andindividual learning preferences of students are taken into ac-count [10, 29, 14], often referred to as personalized learning.“Personalized learning refers to instruction in which the paceof learning and the instructional approach are optimized forthe needs of each learner. Learning objectives, instructionalapproaches, and instructional content (and its sequencing)may all vary based on learner needs. In addition, learningactivities are meaningful and relevant to learners, driven bytheir interests, and often self-initiated.” [27].

A step further from personalized education is the adaptivepersonalized approach that requires both detailed profilingof the student’s personal learning preferences, an extraordi-nary collection and annotation of educational material, andupdating the profile based on the annotation of educationalmaterials [9, 15]. The latter is necessary in order to iden-tify educational material best fitting the student’s profile.Network science can provide important tools that help toachieve these goals, as we explore in this paper.

III. ON BECOMING A SYSTEM

We envision a network science approach serving as thefoundation for real-time personalized adaptive learning. Allcurated educational content contained within the contentmanagement system is presented as a Network of Knowl-edge, that learners can navigate based on prerequisites andcorrelations of topics. Building upon this foundation, thelearning management system must contain a personalizedand individualized recommender system which matcheslearners’ needs (as stored in student profiles) with educa-tional content (stored in the repository of CHUNKs), andpresents each learner with a personalized network as aportion of the Network of Knowledge comprized of all theexisting CHUNKs. Lastly, we envision an adaptive learningenvironment where student profile data and educational con-tent are continuously updated, based on observed patterns oflearners supported by artificial intelligence.

IV. CURATED HEURISTIC USING A NETWORK OFKNOWLEDGE

The proposed CHUNK Learning is a real-time and Adap-tive teaching-learning method for enhanced and personalizededucation. It provides a Curated way of moving througha Network of Knowledge composed of reusable learningobjects joined together by common attributes (i.e., taggedwith competency or skill levels), rather than following thestandard linear or tree-like system of lectures or chapters.CHUNK Learning thus enables the learner to Heuristicallydiscover or learn based on personal background and inter-ests, which we believe will not only enhance the learner’stalents, but will make them a more valuable resource. Thissystem is live at www.ChunkLearning.net, see [1].

How do we achieve it? Our learner’s interests determinehis/her own learning path through the Network of Knowl-edge with individualized learning outcomes. Each studentbenefits differently from the learning experience based onhis/her skills and desires. Simultaneously, the Network ofKnowledge builds on the experiences of the students covertlyguiding learners through the educational materials, muchlike Amazon provides recommendations for buyers. Weachieve this by moving away from interdisciplinary teachingthat transfers methods from one discipline to another, opt-ing instead for a trans-disciplinary teaching approach thatcrosses the boundaries of many disciplines using a diversechoice of teaching tools and software.

V. THE SCIENCE BEHIND CHUNK LEARNING: AMOTIVATIONAL FRAMEWORK

In determining how best to create a platform to supportboth educators and learners, CHUNK Learning focuses onaddressing how to positively affect educational engage-ment as well as stimulate enthusiasm for self-directed life-long learning. Research supports the need for educatorsto effectively connect with learners by addressing learnermotivation. Pink and Brophy identified the importance ofproviding relevance and utility of content as students aremotivated when they understand why they are being askedto learn certain material [23, 8]. It is paramount that thelearning platform focuses on providing learners the answersto “why should I care about this topic.” Additionally, adultlearners are more likely to engage in learning when topicshave intrinsic value and demonstrate importance within thecontext of their lives [19]. Providing context for “how will Iuse this knowledge in my life” ensures learners connect theoverall purpose and goals of instruction.

Personalized learning software respects the reality thatall learners are unique and in order to maintain propermotivation it is necessary to support their different learningstyles and timelines. Utilizing key motivational strategiesto include a variety of content delivery methods and styleshelps ensure learners generate and maintain a positive atti-tude [30]. Additionally, recognizing even the most motivatedlearners, self-directed life-long learners, may expose vari-ables that pose challenges to the pursuit of education. It isnecessary to employ strategies that promote learners’ controlover the delivery of short and intense modules. Respectinglearners’ time and empowering learning-oriented learnersto engage in self-directed study establishes an inclusiveinvitation to education [19].

CHUNK presents learners materials in a variety of formatssupporting multiple pedagogies. This orientation empha-sizes: the use of multiple mental and pedagogical repre-sentations; the promotion of multiple alternative systemsof linkage among knowledge elements; the promotion ofschema assembly (as opposed to the retrieval of prepackagedschemas); the centrality of “cases of application” as a vehiclefor engendering functional conceptual understanding; andthe need for participatory learning, tutorial guidance, and

adjunct support for aiding the management of complexity.This approach is geared towards an active constructionof new knowledge and critical thinking skills, which isespecially learning-enhancing for adult learners who bring alarge array of previous experiences and knowledge to theireducational journey [31]. Similarities in prior experiences orinterests among learners can generate new paths of learningin a network-of-knowledge based system, further deviatingfrom the linear path of the traditional education. We envisionan array of assessments of the CHUNK-generated learningand experiences, relative to the traditional, linear educationalsystem, with the paramount goal of enhancing learning out-comes and guiding the continuous shaping of the CHUNKLearning environment.

VI. EDUCATIONAL MATERIAL AS A NETWORK OFKNOWLEDGE

This research uses a network science [6] approach torepresent educational content that facilitate learning as amap of aggregated materials. This collection of educationalmaterials and the curration that holds it together forms aNetwork of Knowledge formally defined by Cleven as:

A Network of Knowledge is the representation ofeducational material as a (multilayer) network, inwhich educational content builds nodes that arelinked by common attributes in several categories,such as content, prerequisites, instructors, level,etc. Each category of attributes links the nodesin one layer, which provides an ability to filterthe entire collection of material with differentperspectives. [11]

We establish the network using metadata attached tothe educational material. We envision the backbone of thenetwork as a keyword ontology describing relationshipsbetween the content the learners frequently access. However,also included would be other relationships such as contentauthors, prerequisites, media types, and learning methods. Avisualization of the Network of Knowledge thus establishedis depicted in Figure 1.

VII. WHY-HOW-METHODOLOGY-ASSESSMENT

Our work establishes a baseline template that all CHUNKLearning modules follow, displayed in Figure 2. This formatfacilitates learning as an adaptation of Simon’s “Why-What-How” [25] format. For example, in introductory coursesin science, the common practice is to provide a motiva-tion that is easy to follow motivation for the concept-to-be-introduced, with the message that the learner willeventually be using the learned concepts. CHUNK Learn-ing’s “Why-How-Methodology-Assessment” reverses thisprocess. It is top-down teaching by anchoring the concept-to-be-introduced to each learner’s knowledge before intro-ducing the methodology for the new concept. It shows eachlearner how the content is used in that learner’s specific fieldof study, so that it has meaning and context to the learner,before the learner even engages with the new content. This

Fig. 1: CHUNK Network of Knowledge.

is accomplished by having multiple choices for each of thecategory in “Why”, “How”, “Methodology”, and “Assess-ment”, in order to optimize the matching of the content toeach user. Throughout the paper, “user” and “learner” areused interchangeably.

Fig. 2: Why-How-Methodology-Assessment Format.

1) “Why”: Tantalizing the Learner. Learners open a“Why” CHUNKlet to reveal an enticing one-of-its-kindeducational trailer. CHUNK Learning wants to makethe student eager to learn, so all CHUNKs begin witha demonstration on why learning a particular topic isimportant. Much like a movie trailer attracts movie-goers to a movie, the “Why” CHUNKlet attracts anexploratory learner to the CHUNK Learning module,answering the following questions:

• Why is the topic relevant?• Why should students learn the topic?

2) “How”: Applications, Real and Relevant Learnersdive into the “How” CHUNKlet to uncover real andrelevant applications. Here, students discover the an-swer to the often-asked question, “When will I ever usethis in real life?” We also seek to answer the followingquestions:

• How is the topic applied in practice?

• How does the learner validate what he/she alreadyknows?

• How are the learning outcomes tested?• How is new information, anchored to the learner’s

interests, incorporated into the module?• How can the learner apply the acquired

skill/knowledge?3) Methodology: A Variety of Delivery Methods In-

structors carefully curate the “Methodology” CHUN-Klets, guiding students though a variety of personal-ized course materials and delivery methods includingMassive Open Online Courses (MOOCs) and CreativeCommons Licensed resources, as well as instructor-created content. For interactive modules, we envisionthat instructors will follow the “I do it, We do it, Youdo it” model. Their main focus should be on answeringthe following questions:

• What new information and skills will the moduledeliver?

• What activities will the learner be required toperform?

• What learning outcomes will the learner acquire?• What different methodologies could be used to

engage with this new knowledge?4) Assessment: Competency Based Learners jump into

the “Assessment” CHUNKlet to test knowledge of atopic at any point. Assessments are always availablefor every CHUNK. Opportunities for remedial learningare always present. Successful completion results in aCHUNK competency credit.

• What is the competency based framework de-signed around learning objectives needed for eachCUHUNK?

• How should remediation be tested?• How should the post test differ from the pretest?

We encourage instructors to complement this basic structurewith innovative methods of incorporating information andengaging the learners.

VIII. INDIVIDUALIZED SYSTEM

Due to their various academic and professional back-grounds, learners have unique gaps between current knowl-edge and required predefined course knowledge standards.Additionally, individuals have different skills, preferredlearning modes, and motivation to learn. This challengeseducators to expand the one size fits all education method toeffectively engage with learners. CHUNK Learning providesan interactive learning system that is able to identify uniquegaps and then fills those gaps ensuring the same knowledgeoutcomes are met based on student’s uniqueness, such as theones presented in Table I:

In CHUNK Learning, each learner creates a “user profile”tagged with keywords such as the examples from categoriesin Table I, based on the specific user’s background. This pro-file is used to attract CHUNKlets with matching keywords,

Category ExamplesApplication topics Social networks, Cyber, Bitcoin, Biology, Economy,

Neuroscience, Internet, etc.Programming Skills Fortran, C, C++, Python, R, JMP, Matlab

Education GED, A.A., A.S., B.A., B.S., M.A., M.S., Ph.DTraining untrained, trained, need practice, refresher

TABLE I: Sample tags used in the profile set up to supportindividualized experiences

Category ExamplesOccupational Specialty Educator, engineer, etc.Professional Experience Department chair, research team lead, etc.

Interest level Depth, breadth, familiarity,practice, gist

Goals Reinforce something learner knows,expand knowledge, get the gist

Preferred learning modes Videos, PPT, exercises, research papers,simulations, demos

TABLE II: Sample tags used in the profile set up to supportpersonalized experiences

to present individualized learning materials. Each learner’sprofile updates as the learner explores CHUNK Learning. Avideo introducing users to the profile can be found here.

Similarly, each CHUNKlet in the Network of Knowledgeis tagged with keywords such as the examples shown inTable I and more. Once the learners register for particularCHUNKs, the content inside the registered CHUNK getssuggested to users based on the count of keyword matchesbetween the tags of the profile and the ones on the CHUN-Klets of the registered CHUNK. The count of keywordmatches is displayed above the CHUNKlet, allowing the userto make personal decisions and possibly take a scenic routeto complete the CHUNK. The CHUNK Learning’s set of tagand categories grows based on users’ suggestions. A videoof how learners would use this Network of Knowledge canbe found here.

In CHUNK Learning, an author builds the necessaryCHUNKs within a course to ensure all learning objectivesare met at completion of all course’s assessments. TheCHUNK Learning system provides a pre-test capability foreach CHUNK, allowing learners to validate an individualCHUNK by demonstrating competency while bypassing the“why,” “how,” and “methodology” CHUNKlets, and com-pleting the “assessment” CHUNKlet. Learners then engagewith remaining CHUNKs to satisfy the required learningobjectives of the course. The result is an interactive systemthat identifies and satisfies unique gaps of individual learn-ers, respectful of the learner’s time and interests.

IX. PERSONALIZED SYSTEM

CHUNK Learning does more than provide a platform foreducators to individualize instruction for learners, it allowsusers to take control of and be responsible for their learning.The CHUNK Learning system also incorporates informationfrom the learner profile based on the categories and examplesof Table II.

The matching between these categories captured in thelearner profile and the CHUNKlets counts the same waytowards the score displayed above each CHUNKlet. Thedifference is in the type of information it captures inorder to offer a personalized presentation of the educationalmaterials, as well as a personalized path through the content.This is supported by system delivering engaging contenttailored to users’ learning preferences, and relevant to theirspecific interests and realities of their daily and future lives.

By providing a network-based system, CHUNK Learningis able to facilitate an education environment where learnerscan see the connections between the topics captured by theCHUNKs and how they fit together, much like a digitalvisualization of a map of the world. Thus, similarly tonavigating using the map, one can navigate the CHUNKLearning network following the arrows that point fromthe prerequisites of the desired content. As there mightbe multiple ways to get to a destination CHUNK, thisvisualization allows the learners to asses the different pathsto the desired CHUNK or CHUNKs, producing personalizedlearning plans decided by students while guided by thecontent creators.

To see an example of this, consider a learner who likessports and has enrolled in Newton’s First Law of MotionCHUNK, as a refresher of a previous exposure to thetopic. Thus, the CHUNK trailer captured in the “Why”video is presented based on football motivation. This isfollowed by an application to the science of Olympics inthe “How” CHUNKlet. Once the learner was presentedwith these concise, rel and relevant educational videos,he/she is engaged in the learning of the new content, the“Methodology” CHUNKlet. In this case, as this is a refreshof a topic, the user is routed to a Distance Learning typeof environment to reinforce the topic; otherwise, for a newconcept, a classic in class delivery method may be moreeffective. The CHUNK concludes with the assessment.

Fig. 3: An example of a personalized CHUNK for Footballfan, reinforcing Newton’s First Law of Motion

Learners’ curiosity is ignited as they peruse a wide rangeof topics, see the connections between them, and determineareas to engage in deep study, utilizing assessments to meettheir own learning goals.

X. CONCLUSIONS AND FURTHER DIRECTIONS

Traditional education places all the students through thesame topics, at the same time, at the same speed, gener-ally teaching to the average student. But such a systemleaves some students behind, and fails to challenge others.Traditional education also produces fairly similar skilledgraduates, not enhancing the different skills, abilities, andinterests that learners might have. Unlike most establishedsciences, new research fields such as Network Science –an emerging field overseeing traditional computer science,operations research, sociology, and other fields – hint at non-traditional methods of education that serve as a starting pointfor the current proposal.

In this work, we proposed a personalized, adaptive learn-ing platform, called CHUNK Learning. CHUNK Learningbreaks away from the predictable pattern of traditionaleducation models and provides content delivery that respectsthe different capabilities, learning styles, and approaches toproblem-solving of every learner. Students are empoweredby a system that ensures learning is efficient, flexible, and re-spectful of their time. Intense and short educational modules,broken into smaller bits of information, stimulates interestand applicability of topics through the use of ”Why learn”this information. It integrates new information to learner’spre-existing knowledge through a ”How do professionalsin your field use this” demonstration that is personalizedto each learner’s background providing valuable contexton how learners can use that knowledge in their fields ofexpertise. It also provides personalized education based onlearner’s best learning style (both at the instructor level andat topic-to-learn level) and adaptive content and methodol-ogy delivery that is optimized for each learner.

We thus provide an exploratory environment that pro-motes curiosity and creativity, collaboration and collegialityas we search for meaningful ways to bring together andcurate knowledge for generations to come. This vision alsodemonstrates the art of possibility in teaching and learningleading to new partnerships with industry, government, andacademia to support education. It shows the educator’scommitment to providing the best in education that speaksto each learner.

ACKNOWLEDGMENTS

The authors thank the DoD for partially sponsoring thecurrent research.

REFERENCES

[1] Curated heuristic using a network of knowledge, year= 2019, url = www.CHUNKLearning.net, urldate =2019-05-05.

[2] e-classroom: The future of education. https://edtechnyc.org/e-classroom-the-future-of-education/.Accessed: 2018-03-16.

[3] Technology trends driving the future of highereducation. https://www.youtube.com/watch?v=8FlgXB0eeq4. Accessed: 2018-03-16.

[4] Beverly Abbey. Instructional and cognitive impacts ofweb-based education. IGI Global, 1999.

[5] Reka Albert, Hawoong Jeong, and Albert-LaszloBarabasi. Internet: Diameter of the world-wide web.nature, 401(6749):130, 1999.

[6] Albert-Laszlo Barabasi et al. Network science. Cam-bridge university press, 2016.

[7] Lori Breslow, David E Pritchard, Jennifer DeBoer,Glenda S Stump, Andrew D Ho, and Daniel T Seaton.Studying learning in the worldwide classroom re-search into edx’s first mooc. Research & Practice inAssessment, 8:13–25, 2013.

[8] Jere E Brophy. Teaching. International Academy ofEducation and the International Bureau of Education,1999.

[9] Peter Brusilovsky, John Eklund, and Elmar Schwarz.Web-based education for all: a tool for developmentadaptive courseware. Computer networks and ISDNsystems, 30(1-7):291–300, 1998.

[10] Peter Brusilovsky and Christoph Peylo. Adap-tive and intelligent web-based educational systems.International Journal of Artificial Intelligence inEducation (IJAIED), 13:159–172, 2003.

[11] Jan-Daniel Cleven. A multilayer network approach forreal-time adaptive personalized learning, 2018.

[12] Michalis Faloutsos, Petros Faloutsos, and ChristosFaloutsos. On power-law relationships of the in-ternet topology. In ACM SIGCOMM computercommunication review, volume 29, pages 251–262.ACM, 1999.

[13] Alan Feldman, Cliff Konold, Bob Coulter, and BrianConroy. Network science, a decade later: The Internetand classroom learning. Routledge, 2000.

[14] Ralucca Gera, Jessica M Libertini, Jonathan W Ro-ginski, and Anthony Zupancic. The network profilesummary: exploring network science through the lensof student motivation. Journal of Complex Networks,6(3):470–484, 2017.

[15] Peter A Henning, Florian Heberle, Alexander Streicher,Andrea Zielinski, Christian Swertz, Jurgen Bock, Ste-fan Zander, et al. Personalized web learning: Merg-ing open educational resources into adaptive coursesfor higher education. Personalization Approaches inLearning Environments, 55, 2014.

[16] Lilian H Hill. Concept mapping to encourage mean-ingful student learning. Adult Learning, 16(3-4):7–13,2005.

[17] DIAO Lin-lin. A survey of chinese english majors’chunk competence [j]. Journal of Pla University ofForeign Languages, 4, 2004.

[18] Colin McCormack and David Jones. Building aweb-based education system. John Wiley & Sons, Inc.,1997.

[19] S Merriam. The changing landscape of adult learningtheory in comings j, garner b and smith c eds review ofadult learning and literacy: connecting research, policy,and practice, 2004.

[20] George A Miller. The magical number seven, plus orminus two: Some limits on our capacity for processinginformation. Psychological review, 63(2):81, 1956.

[21] Rivka Oxman. Think-maps: teaching design thinking indesign education. Design studies, 25(1):63–91, 2004.

[22] Laura Pappano. The year of the mooc. The New YorkTimes, 2(12):2012, 2012.

[23] Daniel H Pink. Drive: The surprising truth about whatmotivates us. Penguin, 2011.

[24] Jose A Ruiperez-Valiente, Pedro J Munoz-Merino, De-rick Leony, and Carlos Delgado Kloos. Alas-ka: Alearning analytics extension for better understandingthe learning process in the khan academy platform.Computers in Human Behavior, 47:139–148, 2015.

[25] Simon Sinek. Ted talk: How great leaders inspireaction. https://www.ted.com/talks/simon sinek howgreat leaders inspire action. Accessed: 2018-03-16.

[26] Clive Thompson. How khan academy is changing therules of education. Wired Magazine, 126:1–5, 2011.

[27] U.S. Department of Education. National EducationTechnology Plan. https://tech.ed.gov/netp/, January2017.

[28] Daisy Wadhwa. Using artificial intelligence technolo-gies for personalized learning and responsive teaching:A survey (2017).

[29] Daniel S Weld, Eytan Adar, Lydia Chilton, RaphaelHoffmann, Eric Horvitz, Mitchell Koch, James Landay,Christopher H Lin, and Mausam Mausam. Personal-ized online education—a crowdsourcing challenge. InWorkshops at the Twenty-Sixth AAAI Conference onArtificial Intelligence, 2012.

[30] Raymond J Wlodkowski. Enhancing adult motivationto learn: A comprehensive guide for teaching all adults(rev. ed.). San Francisco: Jossey-Basss, 1999.

[31] Susan M Yelich Biniecki and Simone CO Conceicao.Using concept maps to engage adult learners in criticalanalysis. Adult Learning, 27(2):51–59, 2016.

[32] Chi ZHANG, Gang CHEN, Minjuan WANG, andHuimin WANG. Studying on design of small-chunklearning resources in mobile learning [j]. OpenEducation Research, 3, 2009.