International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056
Volume:12Issue:10|Oct 2025 www.irjet.net p-ISSN:2395-0072
Development of personalized Recommendation System for E-Learning Platform
Gajanan Wankhade 1 , Mohak Taywade, Shreyash Meshram
1Gajanan Wankhade Prof, & Guide JD college
2 Mohak Taywade, Student, Police line takli
3 Shreyash Meshram, student , Fetri
Asssistant Professor, Students, Department of Computer and Science and Engineering JD College of Engineering and Management Nagpur
Abstract - In later a long time, e-learning stages such as Coursera, Udemy, edX, and Khan Institute have revolutionized the way individuals procure information and abilities over the globe. With the computerized change of instruction, learners nowadays have get to to an overpowering number of courses, instructional exercises, and learning assets. Be that as it may, exploring this tremendous ocean of substance to discover personalized, important, and skill-appropriate materials remains a major challenge for clients. This inquire about proposes a energetic substance suggestion framework custom fitted particularly for elearning situations, planned to convey personalized learning encounters based on client interface, learning designs, ability levels, and interaction history. The center objective of this framework is to direct learners along effective learning ways by prescribing substance such as courses, instructional exercises, video addresses, andperusing materials. These proposals are not as it were based on what clients have locked in with already, but moreover consider how well they have performed, their favored modes of learning (video, content, intelligently works out), and their career or learning objectives. By analyzing behavioral information and execution measurements, the framework points to improve learning productivity and inspiration through focused on suggestions.
Key Words: Machine Learning, Mobile Application Development, Data Collection, Model Optimization, Agricultural Technology, AI Integration, Future Challenges.
1.INTRODUCTION
The fast development of online instruction has definitely changedhowpeoplelearn,upskill,andmoveovercareers.Elearning platforms such asCoursera, Udemy, edX, and the Khan Institute have become central to this advancement, offeringawiderangeofcoursesacrossdisciplines,openand oftenatalowcost.Bethatasitmay,withthegiganticdeluge of instructional substance, learners are regularly overpowered with choices, making it troublesome to recognize the foremost significant, skill-appropriate, and personalizedlearningmaterials..
1.1 Related Work
Recommendation frameworks have been a conspicuous zone of investigate in both scholastic and commercial spaces,playingabasicpartinpersonalizationforstagessuch as Netflix, Amazon, and Spotify. Within the setting of instruction, a few e-learning stages have coordinated recommender frameworks to upgrade client engagement and learning results. This segment presents a survey of existing strategies, advances, and restrictions recommendationsystems.
1.2 Personalized Learning Path Generation
Ratherthangivingisolatedrecommendations,thesystem builds a progressive learning path. This is achieved by: • Mappingprerequisitesanddependenciesbetweencourses• Identifyinglearner’sskillgapsusingtestresultsandskipped topics • Sequencing content from beginner to advanced levelswithinatopic
2. DEPLOYMENT PLAN & FLOW DIAGRAM
Thearrangementoftheproposedcontentrecommendation system follows a modular microservice architecture to ensure scalability, performance, and effectiveness. The framework is composed of key components including a frontend interface, backend APIs, the recommendation engine, databases, and machine learning model management.ThefrontendisbuiltusingReact.jsorVue.js and deployed as a static web application, typically served throughNGINXorplatformslikeAWSS3withCloudFront for efficient content delivery and fast loading times. This interfaceinteractsdirectlywiththebackendtofetchcourse recommendations, display user dashboards, and manage feedbacksubmissions.
3. CONCLUSIONS
Theproposedcontentrecommendationsystempresentsa dynamicandadaptivesolutiontothegrowingdemandfor personalized learning experiences in e-learning environments. By integrating content-based filtering, collaborative filtering, clustering techniques, and Natural
International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056
Volume:12Issue:10|Oct 2025 www.irjet.net p-ISSN:2395-0072
LanguageProcessing,thesystemdeliversintelligentcourse suggestionstailoredtoeachlearner’shistory,preferences, and goals. Through structured learning paths, users are guidedfromfoundationalconceptstoadvancedtopicsina coherent,measurableprogression.Thehybridapproachnot onlyimprovesrecommendationaccuracybutalsosupports learners in overcoming knowledge gaps, maintaining motivation, and achieving educational outcomes more efficiently. The evaluation of the system demonstrated promisingresultsacrossvariousmetricssuchasprecision, recall,NDCG,anduserengagement.Simulatedtestsindicated asignificantincreaseincoursecompletionratesandlearner satisfaction among users who received personalized recommendations. The use of feedback loops and skill mappingcontributedtocontinuousmodelrefinementand context-awaresuggestions,makingthesystemresponsiveto evolving learner behaviors and interests. Despite its strengths,thecurrentsystemwasevaluatedinacontrolled environment with synthetic data. For broader validation, real-world deployment and A/B testing on platforms like Coursera, Udemy, or a university LMS would be essential. Additionally,long-termlearnertrackingwouldallowbetter understanding of content retention and career outcomes. Looking forward, future enhancements may include the integrationofreinforcementlearningtooptimizelong-term learning to trajectories, voice-based interaction for accessibility,multilingual supporttocaterglobal learners, and cross-platform synchronization. Furthermore, help incorporating psychological models of learning and motivationcouldmakethesystemevenmorehumancentric.
As e-learning continues to expand, intelligent recommendation systems like this will play a vital role in shaping the future of education by making it more personalized, inclusive, and effective. VIII. LIMITATIONS
Despite the strong architecture and performance of the proposed content recommendation system for elearning platforms, several limitations exist that must be acknowledged. First, the evaluation of the system was conductedprimarilyonasyntheticdatasetandcontrolled environment, which may not fully capture the diversity, unpredictability, and behavioral complexity of real-world learners.Assuch,actualperformancemetricsmightdiffer oncedeployedatscaleinalivelearningmanagementsystem (LMS).Additionally,thesystemcurrentlyfocusesontextual andstructureddatainputssuchascoursedescriptions,quiz scores,andratings,whichmayoverlookothervaluabledata likeforuminteractions,projectsubmissions,andtime-series patterns of study behavior. Another limitation lies in the "cold start" problem, which persists despite hybrid methods new usersand newlyaddedcoursesstill suffer frominsufficientdata,makingearlyrecommendationsless reliable. While clustering and collaborative techniques mitigate this issue, the system requires further enhancements like knowledge graphs or interactive onboarding to make stronger early predictions. Furthermore, the recommendation engine, while personalized, may overfit to a learner’s pastbehaviorand
limit exposure to crossdomain or exploratory learning unless explicitly tuned to encourage diversity. From a computational standpoint, continuous model updates, feedback incorporation, and real time recommendation generation demand significant processing resources and efficientdatapipelines,especiallywithagrowinguserbase. Ifnotmanagedproperly,latencyandscalabilitycouldhinder the user experience. Lastly, the system’s explainability is limited; while effective, it doesn’t yet offer transparent reasoningforwhyspecificcoursesarerecommended,which couldbeimportantinbuildingusertrustandengagementin academicsettings.
REFERENCES
[1] Lampropoulos, G. (2023). Recommender systems in education:Aliteraturereviewandbibliometricanalysis. AdvancesinMobileResearch,3(2),829–8504
[2] Da Silva, F. L., Slodkowski, B. K., da Silva, K. K. A., & Cazella,S.C.(2022).Asystematicliteraturereviewon educational recommender systems for teaching and learning:Researchtrends,limitationsandopportunities. Education and Information Technologies, 28, 3289–3328
[3] Afzaal,H.,Farooque,A.A.,Schumann,A.W.,Hussain,N., Mckenzie-Gopsill,A.,Esau,T.,etal.(2021).Detectionof a potato disease (Early blight) using artificial intelligence. Remote Sens. 13, 4115. doi: 10.3390/rs13030411
[4] Agarwal,A.,Sarkar,A.,Dubey,A.K.(2019).“Computer vision-basedfruitdiseasedetectionandclassification,” in Smart innovations in communication and computationalsciences.Eds.Trivedi,S.T.M.C.,Mishra, K.K.,Misra,A.K.,Kumar,K.K.(AdvancesinIntelligent SystemsandComputing,Singapore:Springer),105–115. doi:10.1007/978-981-13-2414-7_11
[5] Shojaei, A. (2024). A scoping review of reinforcement learning in education. Computers & Education: AI, 100016
[6] Li, Y., Liu, K., Satapathy, R., Wang, S., & Cambria, E. (2023). Recent Developments in Recommender Systems:ASurvey.arXivpreprint
[7] Aminuddin, N. F., Tukiran, Z., Joret, A., Tomari, R., Morsin,M.(2022).Animproveddeeplearningmodelof chilli disease recognition with a small dataset. Int. J. Advanced Comput. Sci. Appl. (IJACSA) 13. doi: 10.14569/IJACSA.2022.0130750