Source-linked AI summary
An Ontology-based Adaptive Personalized E-learning System, Assisted by Software Agents on Cloud Storage
Monika Rani, Riju Nayak, O. P. Vyas
TL;DR
The paper addresses limited semantic-web integration and adaptive personalization for changing learner behavior. It proposes an ontology-driven e-learning system using the Felder-Silverman model, monitoring agents, and cloud storage. The system adapts content and learning-style preferences to behavior, with evaluation scores ranging from 3.45 to 3.87 and a computer-science content boundary.
Problem
Semantic-web e-learning requires personalized, adaptive learning that addresses changing learner behavior rather than only delivering content.
Method
The system combines domain ontologies, the Felder-Silverman learning-style model, JADE agents, and cloud-hosted resources to monitor and modify learner preferences.
Results
The system adapts challenges and alternatives to active, reflexive, and changing learner behavior, while evaluation averages range from 3.45 to 3.87 across five dimensions.
Takeaways & Limitations
Adaptive monitoring lets the system adjust learning options as learners’ behavior changes, supporting personalization within the proposed e-learning environment.
Takeaways & Limitations
The course ontology currently covers only computer-science e-learning content, and the evaluation notes limited samples and possible survey omissions.
Abstract
from arXiv · showhide
E-learning and online education have made great strides in the recent past. It has moved from a knowledge transfer model to a highly intellect, swift and interactive proposition capable of advanced decision-making abilities. Two challenges have been observed during the exploration of recent developments in e-learning. Firstly, to incorporate e-learning systems effectively in the evolving semantic web environment and secondly, to realize adaptive personalization according to the learner's changing behavior. An ontology-driven system has proposed to implement the Felder-Silverman learning style model in addition to the learning contents, to validate its integration with the semantic web environment. Software agents are employed to monitor the learner's actual learning style and modify them accordingly. The learner's learning style and their modifications are made within the proposed e-learning system. Cloud storage is used as the primary back-end in order to maintain the ontology, databases and other required server resources. To verify the system, comparisons are made between the information presented and adaptive learning styles of the learner along with actions of agents according to learners' behavior. Finally, various conclusions are drawn by exploring the learner's behavior in an adaptive environment for the proposed e-learning system.
1. INTRODUCTION
The paper addresses semantic-web e-learning challenges by combining ontology-based content representation with adaptive personalization that follows learners’ changing styles. Its system uses software agents and cloud storage to support this approach.
- Ontologies address growing e-learning content, learner-specific search, machine-readable knowledge representation, reasoning, and material reuse.
- Agents capture changes in learning style and store them in the ontology to support adaptive learning.
- The system is deployed on DigitalOcean’s remote cloud host to store ontologies, databases, and other server resources.
- The paper evaluates whether presented content and agent actions correspond to learners’ styles and behavior.
2. FOUNDATIONS
The foundations review e-learning readiness, ontology concepts and languages, ontology development, and applications relevant to personalized semantic-web learning. It frames ontologies as formal, explicit, machine-readable knowledge structures supporting interoperability and reuse.
- E-learning research has examined readiness across students, teachers, organizations, healthcare, and industry, often considering training, confidence, infrastructure, access, and motivation.
- An ontology is presented as a formal, explicit specification of a shared conceptualization for representing learner personalization and learning-resource taxonomies.
- Formal, explicit, and shared representations support interoperability, exhaustive knowledge description, and reduced ambiguity in heterogeneous e-learning environments.
- Ontology types range from top-level and mid-level structures to task, domain, and application ontologies with increasingly specific scope.
- OWL is described as a semantic-web standard that supports expressive semantics and automated reasoning, while RDF represents machine-understandable information as subject-predicate-object triples.
An XML-based Ontology Exchange Language
XOL is described as an ontology exchange framework within the paper’s overview of ontology languages.
- XOL is designed as a framework for exchanging ontology definitions.
Simple HTML Ontology Language (SHOE)
This section covers ontology languages, construction, applications, cloud deployment, and multi-agent personalization. It connects semantic representations and agent monitoring to adaptive e-learning content.
- SHOE extends HTML pages with machine-readable semantic knowledge.
- The ontology is built bottom-up from a knowledge-base application so an abstraction can support similar applications.
- Ontology applications discussed include question answering, recommender systems, and personalized course content.
- Cloud storage hosts the application, learner database, user.owl, course.owl, and learning resources.
- The proposed system narrows its main agenda to personalization, with agents specialized in persistent roles.
- The multi-agent environment adds adaptability by monitoring learner activities and modifying the initial Felder-Silverman learning-style assessment.
3. METHODOLOGY
The proposed system combines cloud-hosted ontologies, a Felder-Silverman learning model, and software agents to personalize e-learning according to learner behavior. Its methodology develops learner and course ontologies, uses semantic querying and reasoning tools, and connects agents with the web application.
- System Architecture: The system stores its web application, domain ontology, learning resources, and MySQL authentication database on a DigitalOcean cloud host.The domain ontology is queried to present required learner information and learning content.
- Multi-Agent Architecture: Agents add adaptive personalization by using maintained learner behavior logs to modify personalization beyond the learning style established by the questionnaire.The agents communicate with JSPs or Servlets through the JADE Gateway API.
- Felder-Silverman Learning Model: The Felder-Solomon ILS questionnaire uses 44 compulsory a/b questions to score four learning dimensions from -11 to 11.The dimensions correspond to sensing/intuitive, visual/verbal, active/reflexive, and sequential/global categories.
- Ontology Development: Learner personalization is stored in user.owl, while course materials and modules are organized in course.owl through abstraction from textbooks or online learning resources.The course ontology abstracts detailed learning materials into a satisfactory taxonomy.
- Ontology Development: Protégé, OWL API, and the HermiT reasoner support ontology construction, programming, and consistency checking.DL Query in Manchester Syntax parses ontology data; an example retrieves students attending a lecture.
GatewayAgent
The proposed system uses a GatewayAgent and JADE-based collaborating agents to connect web requests with ontology updates and adaptive personalization. Its workflow deploys ontologies, databases, and agent services on a cloud host.
- GatewayAgent: GatewayAgent acts as the servlet’s dispatcher and main web interface for the application.
- GatewayAgent: A browser POST is handled by the servlet, which invokes the send-message action through the GatewayAgent.
- GatewayAgent: The GatewayAgent forwards extracted recipient and message content through a BlackBoard communication channel and returns PingAgent’s response to the servlet.
- Multi-agent workflow: The Monitor agent receives the session user_id, periodically checks ontology dimensions, and informs the Update agent when changes exceed a threshold.
- Multi-agent workflow: The Update agent applies dimension changes in response to Monitor’s INFORM ACL message, while Sniffer intercepts messages to demonstrate agent communication.
- System deployment: The workflow creates course and user ontologies, deploys Tomcat and MySQL on DigitalOcean, and runs a JADE platform alongside the application.
4. .RESULTS AND DISCUSSION
The results show an adaptive system that presents learning materials and alternatives according to Felder-Silverman dimensions while recording learner choices. Agents use these observed changes to update learner dimensions in the ontology.
- Active-Reflective: Learner choices such as hiding challenges or revealing them are recorded as potential changes to the Active-Reflective dimension.
- Active-Reflective: The system provides regular challenges to active learners and hides them from reflexive learners, while offering alternatives that support adaptation to changing behavior.
- Sensing-Intuition: Sensing learners receive quizzes by default, whereas intuition learners have quizzes hidden but can access them through an alternative option.
- Visual-Verbal: Visual learners primarily receive video content with text explanations available, while verbal learners receive text with an optional video alternative; repeated choices alter dimensions.
- Sequential-Global: Sequential learners receive ordered content with gallery-view alternatives, whereas global learners receive gallery view with content-view alternatives, and these choices are logged.
- Agent adaptation: The Monitor and Update agents communicate dimension changes, and sufficient logged changes are made permanent in the ontology for real-time adaptation.
- Applications: The proposed system is presented as applicable to education and industrial training, with semantic-web integration supporting resource reusability.
5. EVALUATION OF PROPOSED E-LEARNING SYSTEM
The system was evaluated through questionnaire-based dimensional factors, with limitations involving interpretation, domain specificity, sample size, and survey coverage. Average scores were calculated across five evaluation dimensions.
- The course.owl content is currently limited to computer science and may not apply to learners from other domains.
- Learners may interpret questionnaire questions from different perspectives or scenarios.
- The targeted population limits the sample size, and the survey may omit questions relevant to learners.
- Average scores were reported for the five evaluation dimensions in Fig.18.The supplied figure passage identifies the chart as showing average scores.
- The questionnaire evaluation covered learner, instructor, course, technology, and design dimensions.Each question was scored from 1 to 5, with environmental feedback represented through the questionnaire survey.
6. CONCLUSION & FUTURE WORK
The proposed system integrates semantic-web e-learning through domain ontologies, learning-style modeling, and collaborative software agents. Evaluation reported dimension-level scores, while agents monitored and adapted learner preferences in real time.
- The system integrates an e-learning application with the semantic web through domain ontologies.The ontologies support reusable content and expressive concept-hierarchy relations using OWL.
- The course.owl and user.owl ontologies store learning materials and learner personalization based on the Felder-Silverman learning style model.
- The evaluation reported scores of 3.87 for learner, 3.45 for instructor, 3.74 for course, 3.45 for technology, and 3.659 for design dimensions.
- The Felder-Solomon Index of Learning Styles and HermiT reasoner support preference modeling and ontology consistency checking.
- JADE Agent:Updating and Agent:Monitoring detect changing learner patterns and modify preferences through the application log, user.owl, and JadeGateway in real time.