Intelligent and adaptive microservices and neutrosophic-based learning management systems

Intelligent and adaptive microservices and neutrosophic-based learning management systems

4

Haitham A. El-Ghareeb Information Systems Department, Faculty of Computers and Information Sciences, Mansoura University, Mansoura, Egypt

4.1

Introduction

Adaptive eLearning that is supported by intelligent techniques and methods is one of the ways to support personalized eLearning. Adaptive eLearning overcomes many eLearning limitations and solves some of its challenges. This chapter reviewed different adaptive and intelligent eLearning systems that were proposed over a long period of eLearning research. This chapter introduces novel Neutrosophic and Microservicesbased eLearning system to overcome many of the technical and pedagogical eLearning challenges. Technical challenges includes reusability, scalability, integration, and interoperability. New learning model that attempts to solve different challenges is presented. Though there is no single unified learning model that can be the only right model, this chapter is a step toward a better learning model supported by appropriate adaptive and intelligent features. Most of the artificial intelligence (AI) applications have not yet been expanded to or adopted in widely used eLearning systems, especially open-source systems such as Moodle and Sakai. Current intelligent learning management systems (LMSs) are still in their early stages. AI applications need to handle some problems or to be modified before applying them into LMSs, and AI technology also needs to be brought to opensource communities. The presented adaptive eLearning model integrates different intelligent features, mainly Neutrosophic theory within the system to empower the presented model.

4.1.1 Proposed model This chapter presents innovative intelligent features to improve students’ and instructors’ performance and enhance the presented adaptive eLearning model. Neutrosophic theory utilization in different Microservices to present intelligent features is highlighted in different aspects. Presenting adaptive and intelligent features as Microservices

Optimization Theory Based on Neutrosophic and Plithogenic Sets. https://doi.org/10.1016/B978-0-12-819670-0.00004-4 © 2020 Elsevier Inc. All rights reserved.

64

Optimization Theory Based on Neutrosophic and Plithogenic Sets

with standard interfaces will allow different eLearning systems to adopt them, so they will be reusable. This chapter goes as follows: Section 4.2 discusses different eLearning aspects, mainly pedagogical, technical, adaptive, and intelligent. Section 4.3 presents the basic ideas of Neutrosophic Sets and Theory, focusing on what will be utilized in this chapter. Section 4.4 illustrates the importance of moving eLearning systems from monolithic systems to Microservices-based ones. Section 4.5 discusses the proposed adaptive eLearning model that is enabled via the presented intelligent Microservices. Section 4.6 presents novel neutrosophic utilization in empowering adaptive eLearning with intelligent services. Section 4.7 presents the evaluation of the proposed novel adaptive eLearning model and neutrosophic-based intelligent microservices. The chapter ends with a conclusion and references.

4.2

eLearning

eLearning can be thought of as the learning process created by interaction with digitally delivered content, services, and support. eLearning involves intensive usage of Information and Communication Technology to serve, facilitate, and revolutionize the learning process [1–3]. Learning methods include traditional learning (face-to-face), distance learning (complete asynchronous; both time and place independent learning delivery; mainly online), and blended learning that combines instruction-led learning with online learning activities leading to reduced classroom contact hours. Blended learning has the potential to increase student learning while lowering attendance rates compared to equivalent fully online courses [1]. Blended learning is the learning paradigm that attempts to optimize both traditional learning and distance learning advantages while eliminating learning paradigm shortages and challenges. When compared to traditional learning paradigms, blended learning is found to be consistent with the values of traditional learning paradigms adopted in almost all higher education learning institutions for decades, and has the proven potential to enhance both the effectiveness and efficiency of meaningful learning experiences [2].

4.2.1 Learning management systems An LMS is the software that automates the administration of education. An LMS registers students, tracks courses in a catalog, records data from learners, and provides reports to management. An LMS is typically designed to handle courses by multiple publishers and providers. It usually does not include its own authoring capabilities; instead it focuses on managing courses created by a variety of other learning resources. A prototypical LMS is presented in Riad [3]. Technology and the great advancement in recent web technologies and informal learning methods allowed complete education programs and courses to be presented online.

4.2.2 Pedagogical eLearning challenges Pedagogically, educational psychologists agree that students differ in the ways they learn and very few teachers can adapt learning to each student in typical large

Intelligent and adaptive microservices and neutrosophic-based learning management systems

65

classes. Computer-based learning systems are criticized by many researchers for their limited adaptability of teaching actions compared to rich tactics and strategies employed by human expert teachers [4]. Many universities in developing countries started to adopt eLearning by modifying network infrastructure, establishing new labs, providing an internet connection, and purchasing different tools for creating eLearning courses and using different LMSs. However, these modifications and supplements were not enough to ensure successful eLearning outcomes because other important elements for eLearning success were missing such as flexibility of the system, adaptability toward students’ needs, reusability of learning objects (LOs), interoperability between LMSs, and effective and official design of e-Content [5]. Pedagogically, most training methods and technologies produce, at best, trained novices. That is, they introduce facts and concepts to students, present them with relatively simple questions to test this new knowledge, and provide them with a few opportunities to practice using this knowledge in exercises or scenarios. However, becoming proficient requires extensive proactive solving of realistically complex problems in a wide range of situations, combined with coaching and feedback from managers, more experienced peers, or other types of experts [6].

4.2.3 Adaptive eLearning Adaptive learning for students with many different backgrounds, learning styles, and interests is virtually a must. Educational psychologists by and large agree that students differ greatly in the ways they learn and very few teachers or professors can adapt learning to each student in typical large classes, as the costs associated with delivering different instruction for varied learning styles are prohibitive [7]. Benjamin Bloom [7a] showed 35 years ago, as reported in his 2 sigma paper, that almost all students can learn to a level of mastery, given the right learning environment [8]. In Bloom’s experiments, the most successful learning strategy was tutoring. Adaptive eLearning that is supported with intelligent techniques and methods is one way to support tutoring in eLearning, and thus may be the way to solve many of the limitations and challenges of eLearning today. Adaptive eLearning systems would be a good solution for better eLearning. The vast majority of web-enhanced courses rely on LMSs because they are powerful integrated systems that support a number of teachers’ and students’ needs. Though LMSs are doing great job, indeed for every function that a typical LMS performs there is an Adaptive Web-Based Educational System that can do it much better [9]. Adaptivity is the ability to modify eLearning lessons using different parameters and a set of predefined rules. Researchers differentiate slightly between adaptivity and adaptability by thinking about adaptability as the possibility for learners to personalize an eLearning lesson by themselves. These two approaches go from machine centered (adaptivity) to learner centered (adaptability). In practice, it is quite difficult to isolate one from the other due to their close relationship [10, 11]. Adaptive eLearning is often meant to be new or in an early development stage [4]. An adaptive eLearning system is the environment of software modules, which comprises a set of features for adaptivity and adaptability [12].

66

Optimization Theory Based on Neutrosophic and Plithogenic Sets

Important factors for adapting to student needs and desires include the following [7]: l

l

l

l

l

Each student should move at a unique pace: Given all the variations between students backgrounds, interests, and abilities, it is highly desirable to allow each student to move at a unique pace in the learning units. Adaptation should be very frequent: Changes based on occasional exams are inadequate. Learning activities should adapt to each student on a moment-by-moment basis. Students should feel that the adaptive program is responding to them as individuals. Each student should be successful in learning: A major advantage of adaptive variable placing is that the students can continue to learn in a given area until they have learned the material. Almost all learners can succeed and achieve mastery, but some learners need more time and more practice than others. When something is successfully learned, the learner should move on: Often in classroom learning, after a student has learned something, the class continues working on the topic, boring the student. This will not occur in a fully adaptive learning environment. No one should be taught something he or she already knows: By assuring learner competencies, avoiding unneeded instruction, and moving each student forward when ready, students are expected to achieve a major reduction of learning time, but this cannot be verified empirically until there is a full range of computer-based adaptive learning units.

The provision of static learning material will not meet the requirements of students. Adaptive eLearning enables personalizing the learning process to individual learners via adapting some parameters, like identifying, analyzing, and monitoring relevant aspects of instructions, such as different velocities, paths, or strategies of learning. Performance improvements within the learning process can be gained via adaptive eLearning systems [12]. Adaptation and personalization will improve the learning process; therefore, a paradigm shift from the consumption of static learning contents to well-tailored and highly personalized learning sessions is needed. Over recent decades, various types of adaptation systems and possible areas for their applicability have been identified, leading to the emergence of specialized research fields, like adaptive hypermedia systems, computer-aided instruction, computer managed instruction, recommender systems, intelligent tutoring systems, personalized systems of instruction, and many others. Adaptive multimedia systems as an improved learning environment are well documented in the research work of G€ utl et al. [12].

4.2.4 Intelligent eLearning systems AI utilizes programming algorithms to simulate thought processes and reasoning that produce behavior similar to humans. The applications of AI within eLearning can produce the potential of creating realistic environments with which students can interact. The student essentially would interact with the intelligent agents, which in turn perceive changes in the simulated environment. The intelligent agents would then communicate perceived changes in the environment back to the student who in turn makes decisions based upon their own perceptions of the environment. For modern eLearning systems, we would refer to intelligent agents as intelligent services.

Intelligent and adaptive microservices and neutrosophic-based learning management systems

67

Current learning technologies can help create trained novices more efficiently, but they are really not up to the job of creating true experts. For example, multimedia computer-based training (CBT) systems are good at presenting information and then testing factual recall using multiple choice or fill-in-the-blank questions. However, traditional CBT systems are incapable of providing the intelligent, individualized coaching, performance assessment, and feedback that students need to acquire deep expertise [6]. Employing state-of-the-art AI technology in current eLearning systems can bring personalized, adaptive, and intelligent services to both students and educators. Most AI applications have not yet been expanded to or adopted in widely used eLearning systems, especially open-source systems such as Moodle and Sakai. Current intelligent LMSs are still in their early stages, while AI applications need to handle some problems or to be modified before applying them into LMSs, and AI technology needs to be brought to open-source communities [13]. For detailed study on adaptive and intelligent eLearning systems, and how to integrate Web 3.0 and social networks to present personalized and adaptive eLearning systems, the reader is highly encouraged to review [14].

4.3

Neutrosophic theory

Neutrosophic sets have been introduced to the literature by Smarandache to handle incomplete, indeterminate, and inconsistent information [10]. Neutrosophic theory helps in addressing vagueness, inconsistencies, and missing information. These three challenges face intelligent Microservices, and need to be addressed carefully. In neutrosophic sets, indeterminacy is quantified explicitly through a new parameter I. Truth-membership (T), indeterminacy membership (I), and falsity-membership (F) are three independent parameters that are used to define a Neutrosophic Number. For detailed illustration, discussion, and examples of the utilized Neutrosophic Analytic Hierarchy Process that is utilized in the proposed eLearning model, the reader is referred to Nabeeh et al. [11, 15]. nD E o
(4.1) N ¼ x;TN ðxÞ,IN ðxÞ, FN ðxÞ ,x 2 X








x 2 X, TN ðxÞ, IN ðxÞ,FN ðxÞ 2 ½0,1








(4.2)

Deneutrosophy is the process where Neutrosophic scales/numbers are converted to crisp values by applying score functions of s(aij) as illustrated in Eq. (4.3) [16].   s aij ¼ 1 

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð1  Tx Þ2 + ðIx Þ2 + ðFx Þ2 3

(4.3)

Different Neutrosophic Numbers and Sets are available. They include the following, among others: l

l

single-valued neutrosophic number (SVNN); interval valued neutrosophic number (IVNN);

68 l

l

Optimization Theory Based on Neutrosophic and Plithogenic Sets

single-valued neutrosophic sets (SVNSs); and interval valued neutrosophic sets (IVNSs).

Proposed eLearning model utilizes the novel open-source Neutrosophic package presented in El-Ghareeb [16]. Proposed model utilizes SVNN and SVNS.

4.4

Services-based eLearning systems

4.4.1 Monolithic architecture A monolithic application is an application with a single large codebase/repository that offers tens or hundreds of services using different interfaces such as HTML pages, web services, and/or REST services [17]. A monolithic application has most of its functionality within a single process that is componentized with internal layers or libraries [18]. Monolithic applications scale out by cloning the application on multiple servers or multiple virtual machines. Monolithic applications rely heavily on layered architectures. Monolithic applications face many challenges. An intensive review study of eLearning systems convergence from traditional monolithic systems to services-based adaptive and intelligent systems is presented in Riad [3]. This study highlights the reasons and discusses the necessity of discarding monolithic design when dealing with eLearning systems.

4.4.2 Service-oriented architecture Service-oriented architecture (SOA) is a design pattern that presents IT infrastructure and information systems architecture as loosely coupled, fine granular services that can address system requirements once they are presented by either adding new services or modifying existing ones. SOA also addresses enterprises information systems inefficiency by enhancing reusability, thus theoretically shortening information systems development time and effort required. Besides reusability, interoperability and integration are the other main driving forces for adopting SOA in eLearning systems. W3C defines a Service as A Component capable of performing a task. Consumers need or want is satisfied via a service according to a negotiated contract (implicit or explicit) which includes service agreement, function offered, and so on. SOA is the design pattern that utilizes a services concept to achieve architectural advantages. W3C defines SOA as A set of components which can be invoked, and whose interface descriptions can be published and discovered. This definition can be expanded to include the science, art, and practice of building applications, so SOA can be defined as the policies, practices, and frameworks that enable application functionality to be provided and consumed as sets of services published at a granularity relevant to the service consumer. Services can be invoked, published, and discovered, and are abstracted away from the implementation using a single, standards-based form of interface [19].

4.4.3 Microservices architecture Either Microservices is an SOA done right, or it is a totally new architecture. Microservices has learned from SOA mistakes and failures [20–22]. One important rule for

Intelligent and adaptive microservices and neutrosophic-based learning management systems

69

Microservices is data sovereignty. Microservices is considered the next wave of services-based architecture that has moved SOA from promising architecture to a real success. Microservices architecture has moved from being a promising architecture to become a de facto standard in designing and developing distributed applications. Microservices architecture is used by large companies like Amazon, Netflix, and LinkedIn to deploy large applications in the cloud as a set of small services that can be developed, tested, deployed, scaled, operated, and upgraded independently, allowing these companies to gain agility, reduce complexity, and scale their applications in the cloud in a more efficient way [17]. Microservices can be thought of as an architectural style for developing applications as suites of small and independent (micro)services. Each Microservice is built around a business capability, it runs in its own process, and it communicates with the other Microservices in an application through lightweight mechanisms (e.g., HTTP APIs) [23].

4.5

Proposed adaptive eLearning model

This section presents an adaptive eLearning model as a solution to pedagogical eLearning challenges facing students, and an Adaptive Online Lecture as an enabler to instructors to address adaptivity features in eLearning in a new and innovative form.

4.5.1 Model components Proposed adaptive eLearning model components include: adaptive LMS, quality assurance and accreditation project (QAAP) management system, exam management system, and learning content management system (LCMS) as depicted in Fig. 4.1.

Adaptive LMS

QAAP

Exam management

Learning content management

Student learning subsystem

Course specification module

Exam data

Questions

Student learning profile subsystem

Instructor timetable module

Exam application

Learning objects

Fig. 4.1 Adaptive eLearning model components.

70

Optimization Theory Based on Neutrosophic and Plithogenic Sets

The proposed adaptive eLearning model requires the integration of different systems to achieve the required goal. There are four subsystems: l

l

l

l

Adaptive LMS: Responsible for providing the adaptation features to each student via determining the learning road, topics and time required for each student based on performance, learning profile, and learning preferences. Adaptive LMS provides the basic functionalities provided by different LMSs in an adaptive manner. Adaptive LMS contains two subsystems: student learning and student learning profile. Quality assurance and accreditation project (QAAP) management system: An Egyptian National Initiative and Project that is maintained by the Egyptian Ministry of Higher Education, QAAP includes a course specification module and an instructor module. The course specification module focuses on defining and determining course content, learning objectives, and other course resources. The instructor module contains the instructor timetable that will be used to define suitable times for meetings between students and instructors. Exam management system: A blended model of online questions repository and desktop application delivery exam is used to overcome web-based exam systems’ vulnerabilities to cheating. Students will run the desktop application at exam times. The application will retrieve questions from online repositories. Those repositories are maintained by an LCMS. LCMS: This is critical and vital to the success of the proposed model implementation. The LCMS holds questions items, and LOs. The proposed adaptive eLearning model addresses extra needed metadata for questions and LOs to support needed adaptivity features. The LCMS focuses on providing a stand-alone LOs management that can be utilized by different LMSs. Though the LCMS is thought to be part of the LMS, it is best practice to provide it as a stand-alone system for two reasons: support different LMSs and isolate LOs metadata management from LMS.

4.5.2 Model scenarios To make the proposed model clearer, it will be illustrated in words describing what takes place with students in four different scenarios. Scenario 1 presents Student (A) who uses the system for the first time and has not built the learning profile yet. Scenario 2 continues with Student (A) in the learning phase. Scenario 3 presents Student (B) who is currently doing well through learning, and now has an exam due. Scenario 4 presents Student (C) who failed twice before in the exam, and is taking the exam for the third time.

4.5.2.1 Scenario 1: New student Student (A) attempts to log in but, as it is the first time, finds himself forced to register. During registration, the student completes the forms needed to identify students’ learning profiles and preferences. The second time the student logs in; the student learning subsystem tends to retrieve the student learning preferences from the student learning profile subsystem. If it is not complete, the system forces the student to complete it before starting to learn. Otherwise, the student learning system extracts the student information and registered courses, then checks if this student has an exam. Student (A) does not have an exam, so again, the student learning

Intelligent and adaptive microservices and neutrosophic-based learning management systems

71

system checks if the student has a meeting with an instructor. Student (A) does not have a meeting; otherwise the system would have displayed the calendar.

4.5.2.2 Scenario 2: Studying student The student learning system calls up the QAAP course specification module, and acquires the course specifications and list of topics. The student learning system displays to the student the list of topics, which are available or unavailable as a result of the requirements’ prerequisites, so the student can identify his position on the roadmap. The student selects the topics to learn within the rules. The student learning profile is then updated with the topics selected and not selected. The student learning system displays the suitable learning material and a list of recommended learning materials and what previous students learning the same topics have seen learning materials and what other students currently learning the same topic have seen learning materials list. The student learning system qualifies the student to make sure he understands the learning objectives of the topic. If the student is not qualified, then he goes back to the study plan. The student can quit learning at any time and continue later. Then the student learning system checks if the topics that the student has learned form an exam; if yes, the student becomes eligible for an exam. The student can go through the learn via questions (LVQ) experiment. LVQ is a learning method that simulates the exam environment by presenting questions with feedback, so students can measure their readiness for an exam. The main objective of LVQ is helping students define their readiness level, not testing them. Then the student receives an exam date, and the learning process moves to Student (B).

4.5.2.3 Scenario 3: Due exam student Student (B) logs into the system and has a due exam, so the student learning system attempts to initialize the desktop application responsible for the examination process. The exam application retrieves questions from the LCMS questions repository based on the exam ID submitted by the student learning system. The exam application ranks the exam, and updates the student profile with this rank. The next time the student logs in, the student continues learning new topics.

4.5.2.4 Scenario 4: Suspended student Student (C) faces troubles with some topics. The student took the exam twice but did not pass. So, the third time the student logs in, attends the exam, and does not pass, an automatic initiation of the Intelligent Meeting Manager for Suspended Students occurs to arrange a meeting for this student with one of the instructors to help the student. The meeting details with the student’s detailed profile are mailed to the instructor. The next time the student logs in, he finds the system paused and the calendar is displayed, directing the student to a meeting with the instructor. Only the instructor can solve the situation after meeting the student by reactivating the student account after the proper action has been taken and recorded in the study profile. The instructor can illustrate the topic more than once to the student, examines the student orally, in written form, or by whatever method the instructor finds appropriate.

72

Optimization Theory Based on Neutrosophic and Plithogenic Sets

4.5.3 Adaptive features in proposed model Adaptive features in the proposed adaptive eLearning model include the following: l

l

l

l

l

l

l

Building learning profile and identifying learning preferences for each student using different methodologies. Checking student profile and learning preferences before recommending LOs. Allowing the students to choose among the topics to learn within the constraints of the prerequisites (partial control). Capability to arrange meetings for suspended students. Students are given the chance to selfstudy the subjects and attend the exams three times. If the student fails to pass the exam three times, a meeting must be arranged between the instructor and the student to submit a report by the instructor to the student profile, so the student can continue the learning process again in the adaptive way. This sort of blended learning gives strength to the model. Providing the capability to calculate the required time to study a topic. Tracking student’s behavior in the exams and attempting to identify cheating incidences. Integration with different online forum, wiki, and blog services is available to enhance collaboration between students and encourage them to help each other. Facilities to enable online study groups—like chatting applications—are available.

For detailed study on empowering adaptive lectures through activation of intelligent and Web 2.0 technologies related to the proposed model, the reader is referred to El-Ghareeb and Riad [24]. Supporting online lectures with adaptive and intelligent features related to the proposed model is presented in details in Riad et al. [25].

4.6

Proposed intelligent features

The presented adaptive eLearning model sheds light on supporting eLearning with intelligent features. Intelligence can be addressed in different aspects of the proposed model. This section presents detailed design of the intelligent features to improve student’s performance and help instructors through the eLearning process. Those goals can be achieved by applying different technologies available to educational institutions, instructors, and students in an innovative way. Different intelligent services are presented to enable the intelligent features. Generally, an intelligent service is presented for each intelligent feature. Presented intelligent services can be grouped into two categories based on their users. Fig. 4.2 presents the two proposed intelligent services categories: instructor intelligent services and student intelligent services.

4.6.1 Proposed intelligent microservices From the adaptive eLearning model, intelligent features are as follows: l

l

Students are grouped. Each group is delimited by the same start date. Students who do not catch this start date are delayed to the next group, which is 15 days later. Crawlers keep searching the Internet for newly and updated LOs; in addition, instructors add different resources to the LOs repository. Intelligent learning objects classifier is used to classify found LOs.

Intelligent and adaptive microservices and neutrosophic-based learning management systems

73

Intelligent LO classifier

Intelligent online lecture LOs advisor Instructor intelligent services

Proposed intelligent eLearning services

Intelligent student tracker

Intelligent cheat depressor

Intelligent study plan advisor

Intelligent time-to-learn topic calculation Student intelligent services

Intelligent LOs recommender

Intelligent agenda study time planner Intelligent meeting manager for suspended students

Fig. 4.2 Proposed intelligent eLearning features. l

l

l

l

Learning goals are identified by instructors. Based on these learning goals, instructors define learning paths. Instructors can create some branching at certain points to give students the flexibility to customize their learning paths. The Intelligent study plan advisor helps students at those points. The intelligent time-to-learn topic calculation is the service that is used to advise students about the time needed to study a certain topic. Based on students’ study time of previous topics and the available LOs for this topic, this service can intelligently advise students about the study time issue. Students attend one or more adaptive online lectures within the same learning goal. Adaptive online lectures make use of the intelligent online lecture LOs advisor to recommend LOs for the instructor to use during the lecture, based on the students’ learning profiles. The intelligent LOs recommender is the intelligent service that will recommend LOs for students based on their learning profiles. The recommended LOs list is approved by the instructor and reordered based on the students’ preferences.

74 l

l

l

l

l

Optimization Theory Based on Neutrosophic and Plithogenic Sets

The intelligent agenda study time planner is used to organize the students’ timetables and organize their different activities by connecting them to different activities available in the university based on their preferences through announcements. The intelligent student tracker service is used to track the students’ performances during the online learning journey, and to verify the completeness of students’ learning profiles. Peaks and performance degradation in students’ performances need to be recorded and studied. The learning path is marked by different learning checkpoints. At each checkpoint, students attend an online exam. Those who pass will continue the learning path. Those who fail will have to re-attend the exam within 4 days. If they fail again, they will have to re-attend the exam within 2 days. If they do not pass the third time, they are suspended. The intelligent meeting manager for suspended students service is responsible for managing a meeting between an instructor and the suspended student to handle the learning issues that prevent a student from keeping up with their peer group. Suspended students drop behind their group. The intelligent cheat depressor service focuses on utilizing intelligent techniques in prohibiting students from cheating. When combining the intelligent student tracking service and the intelligent cheat depressor service, cheating instances might be identified.

4.6.2 Instructor intelligent microservices 4.6.2.1 Intelligent LOs classifier Different intelligent techniques can be utilized in classifying LOs based on LO type. Classifying multimedia-based LOs can be via metadata, tags, and annotations, while classifying text-based LOs can be done through accessing and analyzing content. Text classification or categorization is the process of organizing information logically. It can be used in many fields such as document retrieval, routing, and clustering. Document classification tasks can be divided into two sorts: supervised document classification where some external mechanism—such as human feedback—provides information on the correct classification for documents. The second sort is unsupervised document classification, where the classification must be done entirely without reference to external information. The presented intelligent LOs classifier utilizes two of the supervised document classification algorithms: the Naive Bayes Classifier and Term Frequency-Inverse Document Frequency (TF-IDF) algorithms. Both belong to probabilistic classifiers. Intelligent LOs classifier microservice will be covered in a later chapter due to their extensive details.

4.6.2.2 Intelligent adaptive online lecture LOs advisor specifications The intelligent online lecture LOs advisor accesses students’ profiles and learning preferences side by side with data from previous online lectures and course specification data. This service provides the instructor with a recommended list of LOs based on the attending students. This list can be used during the lecture. Table 4.1 presents intelligent online lecture LOs advisor specifications. Algorithm 4.1 presents a detailed algorithm specification highlighting Neutrosophic utilization for this intelligent Microservice.

Intelligent and adaptive microservices and neutrosophic-based learning management systems

75

Table 4.1 Intelligent adaptive online lecture LOs advisor specifications.

Input Student preferences

Related LOs specifications Learning topics data Previous online lectures data

Different learning preferences that identify the student learning behavior are stored. Those preferences are considered for identifying different study plans LOs satisfy students’ classes by percentage. The more available LOs that match students’ preferences, the more this topic is recommended for teaching Data about courses and topics to be pedagogically used in learning scenarios LOs that were used by previous instructors during online lectures for the same topic and students’ feedback for those LOs are important data for this recommendation process

Processing By assigning different weights to the different inputs, Neutrosophic sets are used to generate a weighted list summary report. The intelligent advisor does the following: Identify LOs presented at previous lectures Classify attending students to one of the learning styles Check the LOs specifications and metadata Identify the most suitable LOs to use with attending students l

l

l

l

Output Recommendation report

The instructor can use this report to identify LOs suitability

Algorithm 4.1 Intelligent online LOs advisor algorithm 1: Instructor initiates adaptive online lecture preparation 2: Extract LOs related to the learning topic 3: Extract LOs utilized by previous instructors and students’ feedback 4: Extract instructor recommended LOs for topic 5: Extract student preferences and learning class 6: Extract related LOs specs 7: for all LOs do 8: if instructor previously assigned T, I, F weights to LO then 9: assign T, I, F to instructor recommended LOs for topic 10: end if 11: if LO was utilized by previous instructor then 12: assign T, I, F to LO based on previous instructor evaluation 13: end if 14: if LO was utilized in previous lectures then 15: assign T, I, F to LO based on previous students’ feedback (cumulative)

76

Optimization Theory Based on Neutrosophic and Plithogenic Sets

16: 17: 18: 19: 20: 21: 22: 23: 24:

end if assign T, I, F to LO based on LO matching with student preference end for for all LOs do Deneutrosophy LO and calculate its relevancy to the learning topic Deneutrosophy LO and calculate its similarity to student learning preferences end for Sort different LOs based on their similarity to students Return sorted list of recommended LOs

4.6.2.3 Intelligent student tracker This service tracks student behavior and ensures that there is a complete learning profile that helps the system to identify the students learning preferences and styles all over the system. The learning style is the individual’s characteristic ways of processing information and behaving in learning situations. Knowledge of learning styles can help instructors better understand learners and has important implications for program planning, teaching, and learning.

4.6.2.4 Intelligent cheat depressor The intelligent cheat depressor service tracks students’ behavior in exams and records both: students’ marks, and exam times, trying to identify peaks in marks. Though this service does not detect incidents of cheating for certain; it is used as an indicator to the

Table 4.2 Intelligent cheat depressor Microservice specifications.

Input Student’s previous exam data

Student’s latest exam data

Data include time consumed by student at each exam, type of exam, and mark scored at previous exams, besides different other exam metadata, like number of exam questions, each question difficulty level, etc. Data include time consumed by student at each exam, type of exam, and mark scored at this exam, besides different other exam metadata, like number of exam questions, each question difficulty, etc.

Processing Utilizing Neutrosophic theory to calculate cheating susceptibility. Membership values are adjusted to track and allow the increase of a student’s performance when getting better; however, peak changes are definitely identified

Output Informing instructor to investigate the case when needed

Intelligent and adaptive microservices and neutrosophic-based learning management systems

77

instructor to track certain students. Table 4.2 presents the details of this intelligent Microservice.

4.6.3 Student intelligent microservices 4.6.3.1 Intelligent study plan advisor The intelligent study plan advisor is an intelligent advisory service used to help students identify the appropriate study plans by: l

l

identifying older study plans; and identifying study plans of colleges.

Students differ in their learning behavior and learning preferences. The intelligent study plan advisor service considers different students as classes based on their learning preferences. Table 4.3 presents the intelligent study plan advisor specifications. In this Microservice, instructors identify branching capabilities in the learning path where students can have the opportunity to study a learning topic. Algorithm 4.2 Table 4.3 Intelligent study plan advisor Microservice specifications.

Input Student preferences

Learning class

Study plans for previous students Study plan for colleges

The proposed model stores different learning preferences that identify student learning behavior. Those preferences are considered for identifying different study plans. Students are grouped into classes to ease educational tasks. Classes include auditory, visual, and other classes that are discussed in detail in the learning profile section. Students need to take a closer look of previous instructor plans, grades that students scored by following certain plans, and other data What students in the same groups are studying now

Processing To generate the recommended study plan, the system utilizes Neutrosophic theory and sets as follows: Identify the class to which the students belong Check the branching decision that is assigned by instructor for that class, and double the weights of this decision Check the average of branching decisions taken by students in the same class Ranks recommendations from top-down based on the generated weights l

l

l

l

Output Recommended study plan Study plan for colleges as information

Recommended choice to take in the study plan Display hints on what colleges allow students to study, so student is free to follow their path

78

Optimization Theory Based on Neutrosophic and Plithogenic Sets

Algorithm 4.2 Intelligent study plan advisor algorithm 1: 2: 3: 4: 5: 6: 7: 8:

Student initiates intelligent study plan advisor microservice Extract study plans for previous students Extract instructor recommendations for student class Extract student preferences and learning class Identify student class for all Topics to learn do if instructor previously assigned T, I, F weights to topics to learn then assign T, I, F to instructor recommended next topic to learn for student class 9: end if 10: if Topic was studied before by previous students then 11: assign T, I, F to topic learned by elder students in the same student class 12: end if 13: end for 14: Utilize Neutrosophic rules engine to calculate top-three recommendations of next topics to learn 15: Return three recommendation lists: instructor, previous students, and colleges

presents detailed intelligent study plan advisor implementation specifications utilizing Neutrosophic Sets and Theory.

4.6.3.2 Intelligent time-to-learn topic calculation The intelligent time-to-learn topic calculation is an intelligent Microservice that helps students identify time needed to learn a certain topic. From a study time point of view, the time needed to study a topic is the summation of the time needed to study LOs composing this topic. The instructors define the learning time for each LO as one of the LOs educational metadata attributes. The system can identify learning time variances between instructors’ identified learning time and the students’ actual consumed learning time through tracking students. This time can help students estimate the time needed to finish studying. Table 4.4 presents the intelligent time-to-learn topic calculation specifications.

4.6.3.3 Intelligent LOs recommender The intelligent LOs recommender is the Microservice that aims to find the most pedagogically suitable LO for helping students learn a topic, then personalizing the recommended list based on students’ preferences. Thus, the intelligent LOs recommender must analyze newly introduced LOs efficiently, then store information about them for further processing and ordering to each student. From high-level view, the intelligent LOs recommender executes through two phases: the LOs finding, gathering, and analyzing phase, and the intelligent personalized supervised LOs recommendation phase. The

Intelligent and adaptive microservices and neutrosophic-based learning management systems

79

Table 4.4 Intelligent time-to-learn topic calculation Microservice specifications.

Input Instructor defined learning time Student learning time shift

LOs author defines learning time for each LO. Later, different instructors can identify learning times for the same LO to match students’ skills Tracking the students’ learning progress helped the system to calculate the time-to-learn shift between the defined time and the student actual time to learn. Average time-to-learn calculation will be presented

Processing To calculate total time-to-learn topic for student, system utilizes Neutrosophic Sets and Theory in the following process: System identifies the LOs list the student must learn to finish the topic based on instructors’ directions System identifies the time shift between instructor’s identified learning time and the actual time taken by student. Such an entry is identified over time through tracking student System estimates time needed for learning each LO, and for all LOs forming topic, system calculates timed needed to learn that topic l

l

l

Output Time-to-learn topic

Time estimated for student to learn certain topic

intelligent LOs recommender Microservice will be covered in Chapter 9 due to its extensive details.

4.6.3.4 Intelligent agenda study time planner This helps students identify study times, and integrate activities with their agenda to improve performance. It uses the study time shift estimated via Neutrosophic theory between the instructor LO study time and student actual study time, and can intelligently suggest time needed for students to finish their studies. In addition, it integrates different activities in the university within students’ timetable based on students’ preferences. It presents students’ timetables that combine lecture times, study times, and activity times, so they are personalized for each student. Table 4.5 presents the intelligent agenda study time planner specifications.

4.6.3.5 Intelligent meeting manager for suspended students Students who fail three times when taking an exam are suspended from accessing the system. Not being able to pass the exam after three attempts indicates that there are some pedagogical issues that need taking care of. Suspended students must meet one of the instructors to help them identify and work on solving challenges. Identifying the

80

Optimization Theory Based on Neutrosophic and Plithogenic Sets

Table 4.5 Intelligent agenda study time planner Microservice specifications.

Input Student preferences Related LOs specifications Study time shift

Model stores different learning preferences that identify student learning behavior. In addition, students register in their preferred activities Specifications of LOs those students will study, including instructoridentified study time Neutrosophic-based estimate of the time shift between actual study time identified for each LO and estimation of the actual time the student needs to study this LO

Processing Intelligence in processing takes place in different activities, mainly when conflicts, vagueness, data incompleteness, or inconsistencies occur. The system can resolve conflicts using Neutrosophic Sets and Theory as follows: The system identifies LOs list that student has to study, assigning them the highest weight value The system identifies activities available this week that match student’s interests The system identifies lecture times, and assigns them the highest weight value The system identifies student free time The system attempts to suggest a weekly agenda for student to satisfy all of the above. When conflicts, vagueness, data incompleteness, or inconsistencies occur, Neutrosophic theory is used to assign different T, I, F weights for different activities, and to calculate the importance of each entry. Highest priorities are identified and override low-priority activities l

l

l

l

l

Output Personalized agenda

Personalized agenda for each student that combines lecture times, suggested study times for LOs, and activity times

time for suspended students to meet instructors is an intelligent process that utilizes Neutrosophic Sets and Theory to reach the most suitable time for both students and instructors. Table 4.6 presents specifications of the intelligent meeting manager for suspended students. Suspended students cannot access the system until they are reactivated by the instructor after the meeting.

4.7

Evaluation

Preparing a computer networks course and presenting it to students in the form of the presented adaptive eLearning models and experiencing it resulted in the following satisfaction measures. Table 4.7 presents a summary of students’ opinions about

Intelligent and adaptive microservices and neutrosophic-based learning management systems

81

Table 4.6 Intelligent meeting manager for suspended students Microservice specifications.

Input Student timetable Instructors to meet

Proposed model extracts suitable student meeting times Different instructors can support the same course. Students are able to give priorities for different instructors

Processing By finding matches between students’ available time and instructors’ available time, proposed meeting times are presented. Three different proposed meeting times are presented, and wait for instructors’ approval in order. Arranging meetings faces challenges, especially when there are no free times available. When this is the case, the system needs to break some time constraints using Neutrosophic Sets and Theory to identify what time constraint to break. Instructors must approve meetings before they are sent to the student. Neutrosophic theory is perfect here for calculating and formulating indeterminacy for different schedules that might take place because of unforeseen problems

Output Proposed meeting time

Proposed meeting is approved by instructor. If approved, student is informed of this meeting and now the instructor has full control on the student’s status. The instructor can reactivate the student, make him access LOs, and attend exams if needed

presented features and how they evaluate the need for these and their performance. The presented adaptive eLearning model was tested on a sample of 10 students. Table 4.8 presents a summary of instructors’ thoughts about presented features and how they evaluate the need for these and their performance and behavior. Table 4.7 Summary of students’ evaluation of presented eLearning model features.

Feature Learning preferences Learning profile Customizing course Separate groups Exams check points

Strongly agree (%)

Agree (%)

Neutral (%)

Disagree (%)

Strongly disagree (%)

90

10

–

–

–

85 80

15 10

– 10

– –

– –

30 30

10 10

10 10

25 30

25 20 Continued

82

Optimization Theory Based on Neutrosophic and Plithogenic Sets

Table 4.7 Continued

Feature

Strongly agree (%)

Agree (%)

Neutral (%)

Disagree (%)

Strongly disagree (%)

LVQ Video LOs LOs recommender Agenda study plan Study plan advisor Meeting manager

90 90 70 63 65 40

5 5 20 17 30 10

5 5 – 5 – 8

– – 5 8 5 2

– – 5 7 – 40

Table 4.8 Summary of instructors’ evaluation of presented eLearning model features.

Feature Learning preferences Learning profile Customizing course Separate groups Exams check points LVQ Video LOs LOs recommender LOs advisor Cheat depressor Student tracker

Strongly agree (%)

Agree (%)

Neutral (%)

Disagree (%)

Strongly disagree (%)

90

10

–

–

–

90 70

10 20

– 10

– –

– –

30 70

10 15

10 15

25 –

25 –

90 90 70 73 50 80

5 5 20 17 10 20

5 5 10 10 10 –

– – – – 20 –

– – – – 10 –

4.7.1 Comments on evaluation results To evaluate the adaptive eLearning model, the two target categories of the model were students and instructors. Both of them showed interest in the presented adaptive eLearning model and felt that it could enhance the eLearning experience greatly. Students have issues with the repeated exams process, and grouping students in smaller groups. However, they liked the presented adaptive features. Instructors suspected the applicability of the intelligent cheat depressor service; however, they agreed to use it as an indicator, and the final decision remains their decision.

Intelligent and adaptive microservices and neutrosophic-based learning management systems

4.8

83

Conclusion

eLearning is an important part of the future; however, it is facing challenges. One of the eLearning challenges is the absence of current eLearning systems that adaptively and intelligently invoke students’ capabilities. Adaptive eLearning is the solution to exploit unlimited eLearning advantages. Adaptive eLearning that is supported by intelligent methods and techniques, such as Neutrosophic Sets and Theory is a need. Adaptive eLearning that is supported by intelligent techniques is the solution to present efficient and effective learning. This chapter presented an adaptive eLearning model that blends instructor-led education with eLearning capabilities to provide an enhanced eLearning environment as the solution to current eLearning challenges. Presenting adaptive and intelligent features in the form of Microservices with standard interfaces allows different eLearning systems to adopt them, so they will be reusable, and the newly introduced information systems will not have to reinvent the wheel. In addition, wrapping adaptive and intelligent features with standard interfaces will present a separation of interests that help adaptive and intelligent features researchers and developers to focus more on their target, and transfer the responsibility of utilizing these features in different information systems to information systems specialists. Intelligent Microservices were categorized into two categories based on the user of those Microservices: instructor and student Microservices. The instructor intelligent Microservices are: intelligent LO classifier, intelligent online lecture LOs advisor, intelligent student performance tracker, and intelligent cheating depressor. The student intelligent Microservices are: intelligent time-to-learn topic calculation, intelligent study plan advisor, intelligent agenda study time planner, and intelligent meeting manager for suspended students and intelligent LOs recommender. All those intelligent Microservices utilized Neutrosophic Sets and Theory to overcome difficult situations, mainly incomplete, inconsistent, and missing data. Future work includes the focus on what eLearning would look like in a Web 3.0 world, and how it might differ from current eLearning. eLearning 3.0 is the eLearning empowered by Web 3.0 technologies. eLearning 3.0 will have four key drivers: distributed computing, extended smart mobile technology, collaborative intelligent filtering, and 3D visualization and interaction. eLearning 3.0 will cross the boundaries of traditional learning institutions, and there will be an increase in self-organized learning. With cloud computing and increased reliability of data storage and retrieval, the mashup is a viable replacement for the portal. Both mashups and portals are supporting integration of content provided by other websites and presentation. This will lead to less reliance on centralized provision. Mobiles will play a big part in the eLearning 3.0. There will be a need of ubiquitous access to tools, services, and learning resources, including people—peer learning group, subject specialists, and expert support. Collaborative eLearning will be possible in all contexts. eLearning 3.0 will make collaborating across distance much easier. Three-dimensional visualization will become more readily available. Neutrosophic Sets and Theory will take place in all those scenarios.

84

Optimization Theory Based on Neutrosophic and Plithogenic Sets

References [1] C. Dziuban, J. Hartman, T.B. Cavanagh, P.D. Moskal, Blended courses as drivers of institutional transformation, in: Blended Learning Across Disciplines: Models for Implementation, IGI Global, UK, 2011, pp. 17–37. [2] L. Wang, Implementing and promoting blended learning in higher education institutions: comparing different approaches, in: Comparative Blended Learning Practices and Environments, IGI Global, UK, 2010, pp. 70–87. [3] A.M. Riad, Review of E-learning systems convergence from traditional systems to services based adaptive and intelligent systems, J. Converg. Inf. Technol. 4 (2) (2009) 108–131. [4] F. M€odritscher, V.M. Garcia-Barrios, C. G€utl, The past, the present and the future of adaptive E-learning, in: Proceedings of ICL 2004, 2004. [5] N. Matar, S. Khwaldeh, Z. Hunaiti, Adaptive unified E-learning system for supporting better E-learning approach, in: Annual Postgraduate Symposium on the Convergence of Telecommunications Networking and Broadcasting (PGNet 2007), Liverpool John Moores University, Liverpool, UK, 2007, pp. 28–29. [6] J. Ong, S. Ramachandran, Intelligent tutoring systems: using AI to improve training performance and ROI, Netw. Newsl. 19 (6) (2003) 1–6. [7] A. Bork, C.A. Bagley, X. Zhang, Technology and learning: a proposal for adaptive learning tutorial, Michael Allen’s (2008). [7a] S. Benjamin Bloom, The 2 sigma problem: the search for methods of group instruction as effective as one-to-one tutoring, Educ. Res. 13 (6) (1984) 4–16. [8] D.G. Moursund, Introduction to Information and Communication Technology in Education, University of Oregon, Eugene, 2005. [9] P. Brusilovsky, A distributed architecture for adaptive and intelligent learning management systems, in: Workshop “Towards Intelligent Learning Management Systems”, 11th International Conference on Artificial Intelligence in Education, Citeseer, 2003. _ Otay, C. Kahraman, A state-of-the-art review of neutrosophic sets and theory, in: [10] I. _ Otay (Eds.), Fuzzy Multi-Criteria Decision-Making Using Neutrosophic C. Kahraman, I. Sets, Springer International Publishing, Cham, 2019, pp. 3–24. https://doi.org/ 10.1007/978-3-030-00045-5_1. [11] N.A. Nabeeh, F. Smarandache, M. Abdel-Basset, H.A. El-Ghareeb, A. Aboelfetouh, An integrated neutrosophic-TOPSIS approach and its application to personnel selection: a new trend in brain processing and analysis, IEEE Access 7 (2019) 29734–29744, https://doi.org/10.1109/ACCESS.2019.2899841. [12] C. G€utl, V.M. Garcia-Barrios, F. M€odritscher, Adaptation in E-learning environments through the service-based framework and its application for AdeLE, in: E-Learn: World Conference on E-Learning in Corporate, Government, Healthcare, and Higher Education, Association for the Advancement of Computing in Education (AACE), 2004, pp. 1891–1898. [13] D. Wen, F. Lin, Ways and means of employing AI technology in E-learning systems, in: 2008 Eighth IEEE International Conference on Advanced Learning Technologies, IEEE, 2008, pp. 1005–1006. [14] H.A. El-Ghareeb, Third platform and E-learning ecosystem, EMC Knowledge Sharing (2015). Available from: https://education.emc.com/content/dam/dell-emc/documents/ en-us/2015KS_El-Ghareeb-Third_Platform_and_e-Learning_Ecosystem.pdf.

Intelligent and adaptive microservices and neutrosophic-based learning management systems

85

[15] N.A. Nabeeh, M. Abdel-Basset, H.A. El-Ghareeb, A. Aboelfetouh, Neutrosophic multicriteria decision making approach for IoT-based enterprises, IEEE Access 7 (2019) 59559–59574, https://doi.org/10.1109/ACCESS.2019.2908919. [16] H.A. El-Ghareeb, Novel open source Python neutrosophic package, Neutrosophic Sets Syst. 25 (2019) 136. [17] M. Villamizar, O. Garces, H. Castro, M.V. Merino, L. Salamanca, R. Casallas, S. Gil, Evaluating the monolithic and the microservice architecture pattern to deploy web applications in the cloud, in: 2015 10th Computing Colombian Conference (10CCC), 2015, pp. 583–590. https://doi.org/10.1109/ColumbianCC.2015.7333476. [18] Cesardelatorre, .NET Microservices. Architecture for Containerized .NET Applications (Online), 2019. Available from: https://docs.microsoft.com/en-us/dotnet/standard/ microservices-architecture. (Accessed 14 February 2019). [19] C.C. Chang, K. Hsiao, A SOA-based E-learning system for teaching fundamental information management courses, J. Converg. Inf. Technol. 6 (4) (2011) 298–305. [20] O. Zimmermann, Microservices tenets, Comput. Sci. Res. Dev. 32 (3) (2017) 301–310, https://doi.org/10.1007/s00450-016-0337-0. [21] T. Cerny, Aspect-oriented challenges in system integration with microservices, SOA and IoT, Enterp. Inf. Syst. (2018) 1–23, https://doi.org/10.1080/17517575.2018.1462406. [22] T. Cerny, M.J. Donahoo, J. Pechanec, Disambiguation and comparison of SOA, Microservices and self-contained systems, in: Proceedings of the 2017 Research in Adaptive and Convergent Systems, RACS 2017, 2017, pp. 228–235. [23] J. Soldani, D.A. Tamburri, W.-J.V.D. Heuvel, The pains and gains of Microservices: a systematic grey literature review, J. Syst. Softw. 146 (2018) 215–232, https://doi.org/ 10.1016/j.jss.2018.09.082. [24] H. El-Ghareeb, A. Riad, Empowering adaptive lectures through activation of intelligent and Web 2.0 technologies, Int. J. E-Learning 10 (4) (2011) 365–391. [25] A.M. Riad, H.K. El-Minir, H.M. El-Bakry, H.A. El-Ghareeb, Supporting online lectures with adaptive and intelligent features, Adv. Inf. Sci. Serv. Sci. 3 (1) (2011) 47.