https://dspace.univ-guelma.dz/jspui/handle/123456789/32 2026-04-07T03:10:20Z https://dspace.univ-guelma.dz/jspui/handle/123456789/18956 Indexation des mégadonnées basée IoT dans la chaîne d'approvisionnement alimentaire / Indexing Big Data Based on IoT in Food Supply Chain ZIAYA, Ilyas During the industrial revolution, ensuring food safety has become a critical concern, where the primary objective is to improve the quality of life and safety of citizens. This objective is challenged by the complexity of supply chain processes, which involve the collection of big data from multiple stages and actors, as well as inefficiencies in monitoring and traceability mechanisms that rely on technologies such as IoT and ICT. Alongside these technological factors, Blockchain technology has emerged as a promising solution, offering secure, tamper-proof, and transparent data management across supply chain stages. However, scalability issues hinder traceability query efficiency, as searches must be performed sequentially across blocks. This Ph.D thesis addresses this limitation by developing indexing techniques for big data in the food supply chain to optimize search processes and enhance traceability performance. The first contribution integrates the B-tree indexing technique with the blockchain by introducing a modified transaction structure that records fabrication time and computes per-block ranges. The B-tree structure manages these fabrication time ranges together with their corresponding block numbers, which are continuously updated with each new block until they cover the entire blockchain network. This approach enables the system to efficiently locate relevant blocks containing traceability data, allowing queries to be executed only within the identified blocks, thereby optimizing inter-block searches. The second contribution presents a novel blockchain-based traceability system that combines Natural Language Processing (NLP) with B+ tree indexing, where NLP interprets consumer text queries to identify target supply chain stages and the B+ tree index narrows searches to relevant blocks, ensuring accurate and efficient responses. The third contribution proposes the MerkleB+ tree (MB+ tree), a hybrid structure designed to optimize big data transaction management within each block, reducing linear scans while preserving the security guarantees of the Merkle tree. The simulation of these three systems was conducted using the Hyperledger Fabric framework, which supports food supply chain scenarios and enables the creation of a decentralized network based on our specifications. Performance testing under various network configurations showed that the three systems demonstrated significant improvements in query efficiency, providing faster and more reliable access to traceability information, as well as superior performance compared to existing methods 2026-02-18T00:00:00Z https://dspace.univ-guelma.dz/jspui/handle/123456789/18954 A Hybrid Approach for Improving Recommendation Systems CHEMLAL, Maroua Recommender systems have become ubiquitous on the internet. On most e-commerce and digital service platforms, users frequently encounter suggestions such as “customers who liked this product also liked that one.” The primary goal of these systems is to personalize the browsing experience, optimize the conversion rate (visitor → customer), and facilitate the retrieval of relevant information from massive volumes of data. However, despite their effectiveness, recommender systems face several challenges: (1) the lack of data truly adapted to users’ specific needs, (2) user disorientation during the search process, leading to inefficiency, and (3) difficulties in leveraging the richness of multimodal data collected from diverse sources such as social networks, reviews, interaction histories, and contextual information. Recent research has attempted to address these issues by integrating machine learning and deep learning techniques, as well as multi-criteria decision-making methods, to better capture user preferences, context, and external factors. To overcome these limitations, we propose in this work two complementary recommender systems. The first, the Smart Food and Restaurant Advisor (SFRA), employs Multi-Criteria Decision-Making (MCDM) methods to support healthier and more personalized food choices. Unlike traditional approaches that neglect nutritional and lifestyle dimensions, SFRA integrates user profiles enriched with dietary needs, preferences, and geographic data to deliver context-aware recommendations. Our system evaluates and ranks food and restaurant options by considering nutrition, ingredients, and contextual relevance, enabling users to make informed choices that balance taste with health goals. The second system is an innovative multimodal recommendation approach that exploits heterogeneous data sources (user preferences, spatio-temporal context and emotional) and leverages Graph Neural Networks (GNNs) to model complex relationships between users, items, and contextual factors. Furthermore, we introduce a multi-list recommendation mechanism, designed to provide multiple personalized suggestion lists based on different criteria (relevance, diversity, context), thereby enhancing personalization while reducing user disorientation. 2026-02-16T00:00:00Z https://dspace.univ-guelma.dz/jspui/handle/123456789/17410 DeployingArtificialIntelligencetechniquesfor SupportingDecisionsintheBusinessProcess Area AFIFI, Chaima Business Processes(BP)constitutetheheartofInformationSystems(IS)ofmodern organizations. Thus,theyareintensivelyutilized,bothinthemanagementofvarious companies’ resourcesandindecision-makingandstrategicalignmentactivities.The abstract specifications (or models) expressing thebusinesslogicbehindtheBPsarees- sentialconceptualtoolsusefulforvarioustasks,varyingformmodeling,analysis,moni- toring andmaintenance.However,withthespectacularincreaseinthevolumeofdata handled duringthelifecycleofBPs,whichisoftenheterogeneousinnature,conven- tional approachesformodelingandminingBPmodelsprovetobeineffective,hindering decision-making actions. Toovercometheselimitations,inthisthesisweleveragethelatestadvancementsachieved in theAIareainordertoimprovedecisionsupportsystemsinthefieldofBPsmanage- ment.Thefirstcontributionofthisthesisconsistsofaconceptualframework,calledDSS for BP(DSS4BP),whichallowsconstructingaKnowledgeGraph(KG)thatrepresents the datamanipulatedbytheBPsandtheirlinks.TheconstructedKGispoweredbya graphical capsuleneuralnetwork,anditspurposeistoenablepredictiveanalysisoffu- ture activitiesduringtheprogressionofaBP.ThisDSS4BPisbasedontheG-CAPS-NN architecturetrainedtodiscovertheKG-BP.ThisKGexcelsincapturingcomplexde- pendencieswithintheactivityflowscontainedinthedifferentBPsspecifications.Thus, the developedgraphpromotesahighpredictionoffutureeventsandadeepcontextual understanding ofBPvariationsandevolution.Oursecondcontributionisachat-bot named BPforDecisionSupportSystem(BP-DSS3),whichrefinestheGPT-3.5-turbo chat-bottoassistBPmanagedmakingmoreinformeddecisions.ThisBP-DSS3chat-bot leveragesdeeplearningtechniquestoprovidepersonalizedanddomain-specificdecision support.Afterthetrainingphase,itachievesahighlevelofprecisionandaccuracy when managingreal-worldscenarios,suchasAlignmentwithOrganizationalObjectives (AOO)andRiskManagementandContingencyPlanning(RMCP). The experimentsareconductedbasingonreal-worlddata,thetwoproposedframe- workshavedemonstratedsignificantimprovementsintermsofefficiency,adaptability, and performancecomparedtotraditionalapproaches.DSS4BPenablesorganizationsto proactivelyidentifyinefficienciesandpredictfutureoutcomesofdeployedbusinesspro- cesses, whileBP-DSS3significantlyimprovesdecision-makingbyprovidingactionable and domain-specificinformation. 2025-07-08T00:00:00Z https://dspace.univ-guelma.dz/jspui/handle/123456789/16927 Utilisation de l’intelligence artificielle et l’analyse de l’apprentissage dans la perspective d’améliorer la qualité de l’apprentissage à distance Abda, Youssouf Avec l’essor des technologies de l’information et de la communication, les environnementsd’apprentissage en ligne sont devenus un élément essentiel de l’éducation à tous les niveaux.Ils offrent de nombreuses possibilités d’accès à un large éventail de ressources pédagogiques.Cependant, l’amélioration continue de la qualité de ces cours en ligne reste un défi majeur.Malgré l’existence denombreux modèles d’évaluation, la plupart se concentrent exclusivementsur des aspects structurels et organisationnels tels que la qualité du contenu ou des systèmes desoutien, sans intégrer pleinement les résultats de l’apprentissage de l’apprenant, qui constituentun élément essentiel pour évaluer demanière exhaustive l’efficacité des cours en ligne. Cette thèse vise à combler ces lacunes en développant une approche systématique et innovantede l’évaluation et de l’amélioration de la qualité des cours en ligne dans les environnementsinformatiques pour l’apprentissage humain. Par conséquent, l’objectif principal de ce travail de recherche est de prendre en compte lesrésultats d’apprentissage des apprenants dans le processus d’évaluation et d’amélioration de laqualité des cours en ligne, de manière à ce qu’ils puissent être améliorés en permanence. Pource faire, une échelle de qualité est proposée afin de mesurer la qualité des cours en ligne et deproposer des recommandations pertinentes pour leur amélioration. Notre première contribution dans cette recherche consiste à définir un ensemble de critèresvisant àgarantir la qualité des cours en ligne. Ces critères couvrent plusieurs dimensions essentielles,telles que l’engagement cognitif, affectif et académique des apprenants. Pour chaquecritère, nous avons également identifié et caractérisé les données pertinentes. Ce travail de recherchea impliqué une réflexion approfondie sur la manière de transformer des indicateursqualitatifs, souvent subjectifs et difficiles à quantifier, en métriques précises et mesurables.La deuxième contribution principale de notre travail réside dans le développement d’une approchevisant à améliorer la qualité des cours en ligne, en s’appuyant sur l’analyse d’apprentissageet l’intelligence artificielle. L’approche proposée d’amélioration de la qualité des cours en ligne a été validée à l’universitéde Guelma (Algérie) auprès des étudiants de première année en économie, en utilisant la plateformeMoodle. Les résultats ont montré une amélioration satisfaisante du niveau cognitif desapprenants, confirmant ainsi l’efficacité de cette approche pour améliorer la qualité des coursen ligne. Les recommandations proposées ont été bien reçues et appliquées par les étudiants,suggérant qu’elles influencent positivement leur niveau cognitif 2025-04-14T00:00:00Z