Open research questions in Recommender Systems and Techniques

What the literature leaves open

• representations, tasks, knowledge symbolic mechanisms, neural architectures, and fusion patterns. 2. Qualitative synthesis (thematic analysis): studies were grouped by recommendation task, and their neuro-sym- bolic mechanisms were analyzed to highlight method- ological commonalities, distinctive contributions, and recurring research gaps. This combined synthesis approach supports a coherent interpretation of how neuro-symbolic educational recom- mender systems are currently designed and evaluated, while providing a transparent basis for identifying methodological gaps and research opportunities. 1 3Universal Access in the Information Society (2026) 25:104 3.4 Quality assessment To enhance the methodological robustness of this review and to avoid treating all included studies as equally reliable, a structured quality appraisal was conducted. Given the heterogeneity of neuro-symbolic educational recommender systems and the predominance of experimental research in computer science venues, a concise qualitative assessment framework was adopted rather than a numerical scoring model. Each study was evaluated using four core criteria aligned with the objectives of this review: 1. Clarity of the recommendation task and educational context; 2. Transparency of experimental setup (datasets, evaluation metrics, and baseline comparisons); 3. Explicitness of the symbolic component, ensuring that the study implemented a clearly defined symbolic rea- soning mechanism beyond simple graph representation learning; 4. Clarity of the fusion mechanism, verifying whether the interaction between symbolic and neural components was sufficiently described. The results of this appraisal are summarized in Table 3. Overall, the majority of studies demonstrate strong clarity in task definition and explicit symbolic modeling, Q2 Q1 Q4 Q3 ✓ ✓ ? ✓ ✓ ✓ ✓ ? Table 3 Quality appraisal of included studies (4 criteria) REF Study (short) [35] Gong et al. 2020 [36] Gong et al. 2021 [37] Piao 2021 [38] Vedavathi and Kumar 2022 [39] Obeid et al. 2022 [40] Wu et al. 2023 [41] Yang et al. 2023 [42] Gong et al. 2023 [43] Guan et al. 2023 [44] Sun et al. 2024 [45] Xu et al. 2024 [46] Frej et al. 2024 [47] Lin et al. 2024 [48] Jin and Cui 2024 [49] Zhang et al. 2025 [50] Yang et al. 2025 [51] Li and Luo 2025 [52] Zhou and Wang 2025 [53] Guan et al. 2025 [54] Qin 2025 Ren et al. 2026 [55] ✓ indicates that the quality criterion is clearly satisfied; ? indicates partial fulfillment or insufficient reporting in the original study ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ? ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ? ✓ ✓ ? ? ✓ ✓ ? ✓ Page 7 of 20 104 which is consistent with the inclusion criteria adopted in this review. However, variations were observed in the level of experimental transparency, particularly regarding data- set accessibility and detailed implementation settings. In a limited number of cases, the fusion mechanism was only partially specified, making it more challenging to precisely categorize the integration pattern. The quality appraisal informed the interpretation of the results by enabling a more cautious analysis of studies with partially reported experimental settings or unclear fusion mechanisms. Studies with limited transparency or incom- plete reporting were interpreted with greater caution in the discussion, particularly when comparing performance claims or methodological contributions. Conversely, studies with clear experimental protocols and explicit symbolic– neural integration were considered more robust in support- ing the identified trends. This approach ensures that the synthesis reflects variations in study reliability while pre- serving the inclusiveness of the review.

  Toward transparent and adaptive educational recommender systems: a systematic review of neuro-symbolic knowledge graph approaches · 2026 · DOI

• However, automated or dynamic KG construction whether through large language models (LLMs) or real-time learner interaction data remains underexplored, with only four stud- ies adopting such techniques.

  Toward transparent and adaptive educational recommender systems: a systematic review of neuro-symbolic knowledge graph approaches · 2026 · DOI

• hybrid ideological and political education. Electronics, 15(5), 1086. book Libin, Y., Yu, Z., Xiaoyan, C., et al. (2018). LSTM-based personalized context-aware citation recommendation. IEEE Access, 6, 59618–59627. model for Lin, D., Liu, Y., & Wu, Z. (2026). Cross-course 26 AI Reshaping the Classroom: Transformation of Teaching and Learning Models 2026 knowledge tracing for prediction in MOOCs. Electronics, 15(3), 642. student performance Liu, H., Yu, J., Zhang, L., et al. (2022). NeuMF: Predicting anti-cancer drug response through a neural matrix factorization model. Current Bioinformatics, 17(9), 835–847. Najafabadi, K. M., Kritharides, K., Choi, C., et al. (2026). Bridging the implementation gap in AI-powered A systematic review. AI, 7(2), 41. personalized education: Peng, D. (2024). Civic education reform based on learning model. Applied deep reinforcement Mathematics and Nonlinear Sciences, 9(1). Prabhakar, T., Prasad, M., Kumar, G., et al. (2024).

  Large-Scale Classroom-Based Social Recommendation Model Using Heterogeneous Graph Neural Networks and Social Regularisation · 2026 · DOI

• Kim, D., Jung, H., Lee, J., et al. (2026). From vulnerability robustness: Radiation-hard isolation for CFETs. Nuclear Engineering and Technology, 58(5), 104104. to Le, V. H., & Ho, N. T. T. (2026).

  Large-Scale Classroom-Based Social Recommendation Model Using Heterogeneous Graph Neural Networks and Social Regularisation · 2026 · DOI

• European Radiology, 34(12), 7673–7689. ESUR. by Baarir, F. N., Bourekkache, S., Ammari, A., et al. (2025). LSTM-based prediction of emotional patterns from behavioral data. Education and Information Technologies, 31(3), 1–28. in MOOCs Camilleri, P., Watted, A., & Tonna, A. M. (2025). Investigating teachers ’ changing perceptions towards MOOCs technology through acceptance model. Education Sciences, 15(10), 1395. the Cen, Y., Jiang, S., Cai, W., et al. (2025). EGRec: A MOOCs course recommendation model based on graphs. Discover Applied knowledge Sciences, 7(6), 578. Chen, G., He, Y., & Kong, Y. (2025). Research on on recommendation TrustSVD++ of and XGBoost. Electronic Research and Application, 9(3), 342 –349.

  Large-Scale Classroom-Based Social Recommendation Model Using Heterogeneous Graph Neural Networks and Social Regularisation · 2026 · DOI

• the performance of the model. We will observe different values on the validation set by grid search. 𝛼, 𝛽), and draw the sensitivity thermal map. The results of the thermodynamic map presented in Figure 8 will for provide an intuitive basis the selection of hyperparameters and reveal the deep law of the interaction between social signals and structural signals within the model.

  Large-Scale Classroom-Based Social Recommendation Model Using Heterogeneous Graph Neural Networks and Social Regularisation · 2026 · DOI

• aspect-aware dependence on predefined meta-paths (Fu, 2026). embedding with for multiple network courses, reduce to Current research suffers from fragmentation: social recommendation models fail to fully utilise the heterogeneous network structure of MOOCs, whilst in-depth HIN recommendation models integration of complex social relationships.

  Large-Scale Classroom-Based Social Recommendation Model Using Heterogeneous Graph Neural Networks and Social Regularisation · 2026 · DOI

• new stage. Researchers have begun to use GNN to model higher-order connections and complex influence diffusion processes in user social networks. For example, the DiffNet model recursively aggregates the features of users’ social circles through a neural influence diffusion network to capture the social influence hidden in the deep network (Seah et al., 2014). The subsequent DiffNet++ attempts to model both social influence diffusion and interest diffusion in a unified framework. These graph-based methods are able to better capture relational dependencies in non-Euclidean spaces, providing more expressive power for social recommendation (Chen et al., 2026). It is a logical and promising direction to apply the social recommendation paradigm to educational scenarios such as MOOCs, and more and more studies have begun to explore this cross-cutting field in recent years. These studies generally focus on two core tasks: learning resource recommendation and learning peer recommendation. In terms of resource social recommendation, behavior data, such as forum interaction and learning group participation to enhance recommendation (Wang & Ge, 2025). For example, there are systems that analyze the spontaneous interaction of learners their educational on social networks, construct interest portraits, and then recommend relevant informal learning resources (Zibo et al., 2022). In try to use researchers entered has a al., partners learning 2023). Zheng terms of peer recommendation, the goal is to match or learners with potential learning, collaborative groups to promote social knowledge construction, and emotional support, thereby reducing loneliness and dropout rates (Yuan a et three-dimensional to recommend peers based on knowledge relevance, social proximity and technology accessibility (Libin et al., 2018). Prabhakar et al. designed a reciprocal recommendation system for MOOCs to match communication partners that both sides may be interested in based on the profile attributes of learners (Prabhakar et al., 2024).

  Large-Scale Classroom-Based Social Recommendation Model Using Heterogeneous Graph Neural Networks and Social Regularisation · 2026 · DOI

• gradually become an independent and active research social 15 a by and prior social circles such as framework introducing (Jdidou et factorization branch. Its core idea is derived from an intuitive sociological principle that users are more likely to accept and adopt the suggestions of trusted members in their al., 2025). Incorporating social relationship information into the recommendation algorithm is proved to not only improve the accuracy of recommendation, but also effectively alleviate the problem of data sparsity and cold start. Early classical works, the trust-based recommendation proposed by Massa and Avesani, use explicit trust networks among users for propagation and prediction (Simone et al., 2014; Avesani et al., 2024). Golbeck’s TidalTrust model calculates indirect trust through the aggregation of trust paths (Seo & Han, 2010). These pioneering studies have laid the foundation for trust-aware recommendation. Subsequently, the combination of matrix social information has led to a series of important models. The Social Regularization (SoReg) model proposed by Sun et al. encodes the social relationship as a smoothing social regularization term in the objective function to constrain the latent feature vectors of friend users to be as similar as possible (Sun et al., 2022). The TrustSVD model further developed by Chen et al. (2025), innovatively considers both the explicit and implicit effects of user ratings and trust relationships, extends on the SVD++ framework, enriches user representations by modeling implicit feedback from trusted users, and demonstrates superior performance on multiple public data sets (Chen et al., 2025). With the rise of graph neural networks (GNNs), social recommendation new stage. Researchers have begun to use GNN to model higher-order connections and complex influence diffusion processes in user social networks. For example, the DiffNet model recursively aggregates the features of users’ social circles through a neural influence diffusion network to capture the social influence hidden in the deep network (Seah et al., 2014). The subsequent DiffNet++ attempts to model both social influence diffusion and interest diffusion in a unified framework. These graph-based methods are able to better capture relational dependencies in non-Euclidean spaces, providing more expressive power for social recommendation (Chen et al., 2026). It is a logical and promising direction to apply the social recommendation paradigm to educational scenarios such as MOOCs, and more and more studies have begun to explore this cross-cutting field in recent years. These studies generally focus on two core tasks: learning resource recommendation and learning peer recommendation. In terms of resource social recommendation, behavior data, such as forum interaction and learning group participation to enhance recommendation (Wang & Ge, 2025). For example, there are systems that analyze the spontaneous interaction of learners their educational on social networks, construct interest portraits, and then recommend relevant informal learning resources (Zibo et al., 2022). In try to use researchers entered has a al., partners learning 2023). Zheng terms of peer recommendation, the goal is to match or learners with potential learning, collaborative groups to promote social knowledge construction, and emotional support, thereby reducing loneliness and dropout rates (Yuan a et three-dimensional to recommend peers based on knowledge relevance, social proximity and technology accessibility (Libin et al., 2018). Prabhakar et al. designed a reciprocal recommendation system for MOOCs to match communication partners that both sides may be interested in based on the profile attributes of learners (Prabhakar et al., 2024).

  Large-Scale Classroom-Based Social Recommendation Model Using Heterogeneous Graph Neural Networks and Social Regularisation · 2026 · DOI

• the systems mainly traditional recommendation paradigm. Collaborative filtering methods generate recommendations by analyzing groups of learners with similar ratings or learning behaviors. Its basic assumption is that users learning with similar the resources ubiquitous data cold-start problems in MOOCs make the performance of traditional collaborative recommendation methods based on ratings limited, and new learners or new courses lack sufficient historical interaction data to find reliable similar neighbors (Najafabadi et al., 2026). Content-based recommendation focuses on the matching of learning resource metadata and learner profiles, but this method is highly dependent on high-quality and structured resource annotation, which is difficult to guarantee in the open and dynamic MOOCs environment, and easily leads to the lack of diversity and surprise of recommendation results. In order to overcome the shortcomings of a single method, hybrid recommendation strategies trying to combine collaborative have emerged, recommendation, other information sources. For example, some studies have integrated the recommendation process, or used deep learning models from multi-source data (Chen, 2025). representations content-based recognition learning learn style joint into and to Recognizing the key role of social factors in learning, system has recommendation gradually become an independent and active research social 15 a by and prior social circles such as framework introducing (Jdidou et factorization branch. Its core idea is derived from an intuitive sociological principle that users are more likely to accept and adopt the suggestions of trusted members in their al., 2025). Incorporating social relationship information into the recommendation algorithm is proved to not only improve the accuracy of recommendation, but also effectively alleviate the problem of data sparsity and cold start. Early classical works, the trust-based recommendation proposed by Massa and Avesani, use explicit trust networks among users for propagation and prediction (Simone et al., 2014; Avesani et al., 2024). Golbeck’s TidalTrust model calculates indirect trust through the aggregation of trust paths (Seo & Han, 2010). These pioneering studies have laid the foundation for trust-aware recommendation. Subsequently, the combination of matrix social information has led to a series of important models. The Social Regularization (SoReg) model proposed by Sun et al. encodes the social relationship as a smoothing social regularization term in the objective function to constrain the latent feature vectors of friend users to be as similar as possible (Sun et al., 2022). The TrustSVD model further developed by Chen et al.

  Large-Scale Classroom-Based Social Recommendation Model Using Heterogeneous Graph Neural Networks and Social Regularisation · 2026 · DOI

• sparsity and severe like similar al., 2026). However, As a core tool to cope with information overload, personalized has experienced a deep migration from general domain to vertical domain in the past two decades (Camilleri et al., 2025). The education field, especially MOOCs environment, due to its unique goal-oriented, social interaction and data heterogeneity, puts forward special requirements for recommender systems that are different from e-commerce and entertainment educational scenarios. research recommendation the systems mainly traditional recommendation paradigm. Collaborative filtering methods generate recommendations by analyzing groups of learners with similar ratings or learning behaviors. Its basic assumption is that users learning with similar the resources ubiquitous data cold-start problems in MOOCs make the performance of traditional collaborative recommendation methods based on ratings limited, and new learners or new courses lack sufficient historical interaction data to find reliable similar neighbors (Najafabadi et al., 2026). Content-based recommendation focuses on the matching of learning resource metadata and learner profiles, but this method is highly dependent on high-quality and structured resource annotation, which is difficult to guarantee in the open and dynamic MOOCs environment, and easily leads to the lack of diversity and surprise of recommendation results. In order to overcome the shortcomings of a single method, hybrid recommendation strategies trying to combine collaborative have emerged, recommendation, other information sources. For example, some studies have integrated the recommendation process, or used deep learning models from multi-source data (Chen, 2025). representations content-based recognition learning learn style joint into and to Recognizing the key role of social factors in learning, system has recommendation gradually become an independent and active research social 15 a by and prior social circles such as framework introducing (Jdidou et factorization branch. Its core idea is derived from an intuitive sociological principle that users are more likely to accept and adopt the suggestions of trusted members in their al., 2025). Incorporating social relationship information into the recommendation algorithm is proved to not only improve the accuracy of recommendation, but also effectively alleviate the problem of data sparsity and cold start. Early classical works, the trust-based recommendation proposed by Massa and Avesani, use explicit trust networks among users for propagation and prediction (Simone et al., 2014; Avesani et al., 2024). Golbeck’s TidalTrust model calculates indirect trust through the aggregation of trust paths (Seo & Han, 2010). These pioneering studies have laid the foundation for trust-aware recommendation. Subsequently, the combination of matrix social information has led to a series of important models.

  Large-Scale Classroom-Based Social Recommendation Model Using Heterogeneous Graph Neural Networks and Social Regularisation · 2026 · DOI

• educational contexts. In practice, HIN-SR offers a constructing solution viable next-generation socialised intelligent personalised learning support systems, with the potential and completion rates by improving recommendation accuracy social interactions.

  Large-Scale Classroom-Based Social Recommendation Model Using Heterogeneous Graph Neural Networks and Social Regularisation · 2026 · DOI

• large-scale integrates model heterogeneous information networks with social relationship regularisation. The model utilises a heterogeneous graph neural network to learn the representations of entities within the network and (HIN-SR) deeply that 14 Corresponding Author: Jin Lu Shenzhen Polytechnic University, P.R. China ©The Author(s) 2026. Published by BONI FUTURE DIGITAL PUBLISHING CO.,LIMITED.

  Large-Scale Classroom-Based Social Recommendation Model Using Heterogeneous Graph Neural Networks and Social Regularisation · 2026 · DOI

• There are many opportunities to further improve group rec- ommendation systems using HFSDSim. Currently, this study has employed K-Means clustering for group formation and the average aggregation strategy for combining preferences. In future, we can integrate HFSDSim with other cluster- ing algorithms or various deep clustering algorithm to cre- ate better user groups. For more accurate and personalized group recommendation, we can also include the trust factor and leadership in group in the similarity measure. Also, we can explore more group aggregation strategies to better cap- ture and balance individual preferences within the group. Testing the method on larger and more diverse datasets, such as those from e-commerce, social networks, or healthcare, will help prove its usefulness in different real-world settings. Balancing the preferences of the whole group with those of individual members is another important goal to improve fairness and satisfaction. Finally, building user-friendly systems that explain recommendations clearly can increase user trust and acceptance. By working on these areas, future research can make group recommendation systems more effective, fair, and easy to use. Author’s contribution S. Singhal contributed to conceptualization, methodology design, formal analysis, writing—original draft, and writ- ing—review & editing. K. Pal was involved in supervision, critical revision of the methodology, validation of results, writing—review & editing, and project administration. Both authors contributed signifi- cantly to the interpretation of results and provided insightful feedback on the manuscript. All authors have read, revised, and approved the final version of the manuscript for submission. Data availability Data will be shared if required.

  A new hybrid similarity-based framework for effective group recommendation system · 2026 · DOI

• Future work will focus on achieving cross-art style (such as from oil painting to traditional Chinese painting) path transfer capabilities through meta-learning, and exploring the combination of large language models and reinforcement learning to support more open and flexible learning goal interpretation and recommendation.

  <b>Research on Personalized Art Learning Path Recommendation Algorithm Based on Reinforcement Learning</b> · 2026 · DOI

• learning environments using a learner preference tree and a GA[J]. IEEE Trans Learn Technol 2013;6(4):350–63. framework personal for M.L. Littman, Reinforcement learning improves behaviour from evaluative feedback, Nature 521 (2015) 445–451. 10 D. Silver, T. Hubert, J. Schrittwieser, I. Antonoglou, M. Lai, A. Guez, M. Lanctot, L. Sifre, D. Kumaran, T. Graepel, et al., A general reinforcement learning algorithm that masters chess, shogi, and go through self-play, Science 362 (6419) (2018) 1140–1144. A. Kara, I. Dogan, Reinforcement learning approaches for inventory specifying ordering policies of perishable systems, Expert Syst. Appl. 91 (2018) 150–158. X. Wang, Y. Wang, D. Hsu, Y. Wang, Exploration in interactive personalized music recommendation: A reinforcement learning approach, ACM Trans. Multimed. Comput. Comm. Appl. 11 (1) (2014) 22 F. Liu, X. Li, H. Guo, R. Tang, Y. Ye, X. He, End-to-end deep reinforcement learning based recommendation with supervised embedding, in: Proceedings of the 13th ACM International Conference on Web Search and Data Mining, 2020, pp. 384–392. G. Zheng, F. Zhang, Z. Zheng, Y. Xiang, N.J. Yuan, X. Xie, Z.J. Li, DRN: A deep reinforcement learning framework for news recommendation, in: Proceedings of the 27th International Conference on World Wide Web, 2018, pp. 167–176 V. Mnih, K. Kavukcuoglu, D. Silver, A.A. Rusu, J. Veness, M.G. Bellemare, A. Graves, M. Riedmiller, A.K. Fidjeland, G. Ostrovski, et al., Human-level control through deep reinforcement learning, Nature 518 (7540) (2015) 529–533. J.K. Gupta, M. Egorov, M. Kochenderfer, Cooperative multi-agent control using deep reinforcement learning, in: G. Sukthankar, J. Rodriguez Aguilar (Eds.), Autonomous Agents and Multiagent Systems, Springer, Cham, 2017, pp. 66–83 S. Almahdi, S.Y. Yang, An adaptive portfolio trading system: A riskreturn portfolio optimization using recurrent reinforcement learning with expected maximum drawdown, Expert Syst. Appl. 87 (2017) 267–279. P. Gabrielsson, U. Johansson, High-frequency equity index futures trading using recurrent reinforcement learning with candlesticks, in: IEEE Symposium Series on Computational Intelligence, 2015, pp. 734–741. P. Basile, C. Greco, A. Suglia, G. Semeraro, Deep learning and hierarchical reinforcement learning for modeling a conversational recommender system, Intell. Artif. 12 (2) (2018) 125–141. O. Nachum, S.S. Gu, H. Lee, S. Levine, Data-efficient hierarchical reinforcement learning, in: Advances in Neural Information Processing Systems, 2018, pp. 3307–3317. A.S. Vezhnevets, S. Osindero, T. Schaul, N. Heess, M. Jaderberg, D. Silver, K. Kavukcuoglu, FeUdal networks for hierarchical reinforcement learning, in: Proceedings of the 34th International Conference on Machine Learning, 2017, pp. 3540–3549. J. Zhang, B. Hao, B. Chen, C. Li, H. Chen, J.

  Retraction notice: Efficient Course Recommendation using Deep Transformer based Ensembled Attention Model. · 2026 · DOI

• environments, World Wide Web 17 (2) (2014) 271–284. S. Wan, Z. Niu, An e-learning recommendation approach learning resource, based on the self-organization of Knowl.-Based Syst. 160 (2018) 71–87 algorithm adapted in J. Zhang, F. Gu, Y. J. Ji, and J. F. Guo, “Personalized scientific resources technological recommendation based on deep learning,” Journal of Intelligent Fuzzy Systems, vol. 41, no. 2, pp. 2981–2996, 2021. literature and Zhang B, Yang FZ, Fan LN. Query recommendation based on document content user focuses on. Appl Mech Mater 2014;511:385–8 J, Kim for SB. Content-based recommendation systems using multiattribute networks[J]. Expert Syst Appl 2017;89:404–12 Son filtering Ng YK. CBRec: a book recommendation system for children using the matrix factorisation and content-based filtering approaches[J]. Int J Bus Intell Data Min 2020;16(2):129–49. Sakboonyarat S, Tantatsanawong P. Massive open online courses (MOOCs) recommendation modeling using deep learning[C]//. the 2019 23rd international computer science and engineering conference (ICSEC). IEEE; 2019. p. 275–80 In: Proceedings of Verbert K, Manouselis N, Ochoa X. Context-aware recommender systems for learning: a survey and future challenges[J]. IEEE Trans Learn Technol 2012;5(4): 318– 35 Zapata A, Menendez VH, Prieto ME. A hybrid recommender method for learning objects [J]. In: IJCA Proceedings on design and evaluation of digital content for education (DEDCE). 1; 2011. p. 1–7. Salehi M, Kamalabadi IN, Ghoushchi MBG. An effective recommendation learning environments using a learner preference tree and a GA[J]. IEEE Trans Learn Technol 2013;6(4):350–63. framework personal for M.L. Littman, Reinforcement learning improves behaviour from evaluative feedback, Nature 521 (2015) 445–451. 10 D. Silver, T. Hubert, J. Schrittwieser, I. Antonoglou, M. Lai, A. Guez, M. Lanctot, L. Sifre, D. Kumaran, T. Graepel, et al., A general reinforcement learning algorithm that masters chess, shogi, and go through self-play, Science 362 (6419) (2018) 1140–1144. A. Kara, I. Dogan, Reinforcement learning approaches for inventory specifying ordering policies of perishable systems, Expert Syst. Appl. 91 (2018) 150–158. X. Wang, Y. Wang, D. Hsu, Y. Wang, Exploration in interactive personalized music recommendation: A reinforcement learning approach, ACM Trans. Multimed. Comput. Comm. Appl. 11 (1) (2014) 22 F. Liu, X. Li, H. Guo, R. Tang, Y. Ye, X. He, End-to-end deep reinforcement learning based recommendation with supervised embedding, in: Proceedings of the 13th ACM International Conference on Web Search and Data Mining, 2020, pp. 384–392.

  Retraction notice: Efficient Course Recommendation using Deep Transformer based Ensembled Attention Model. · 2026 · DOI

• provide algorithms, which alternative class suggestions. Verbert et al. studied the context-aware recommendation system within the field of technology-enhanced learning. They developed a contextual analysis dimension and framework. They also analysed existing context-aware guidance technologies and proposed future developments. Some specialists have also investigated the technological implementation and use of contextual suggestion. 2.1.4 Hybrid course recommendation method The term "hybrid recommendation" is used to describe the two or more recommendation practise of merging algorithms; involves a content-based this typically, recommendation algorithm and a CF algorithm. To accomplish individualised course suggestion, Zapata et al. employed LMS metadata information, LRM data, and learner characteristics to influence their approach to content filtering, CF, and learner main search. In order to optimise the weight of the course's implicit attributes, Salehi et al. used a genetic algorithm in conjunction with the nearest neighbour CF algorithm to create a learner interest tree from the course's explicit multidimensional attributes and the learner's historical rating of the course.

  Retraction notice: Efficient Course Recommendation using Deep Transformer based Ensembled Attention Model. · 2026 · DOI

• exponential growth of ICT, MOOCs, and Learning Management Systems (LMS), online learning resources have exploded. Due to disparities in cognitive ability and preferences, learners are unable to rapidly identify and select the learning resources in which they are interested and required. Therefore, it is imperative to develop an intelligent model that accurately and efficiently recommends useful and engaging learning resources to students. In this section, we examine extant relevant research in the following fields: (1) Conventional algorithms for course recommendation consist of context-based, content-based, collaborative filtering (CF), and hybrid recommendations.

  Retraction notice: Efficient Course Recommendation using Deep Transformer based Ensembled Attention Model. · 2026 · DOI

• model, wherein learning objects are simulated as intelligent object entities using the self-organization theory, and the objects in these simulations naturally interact with one another. self-organization-based things, which Despite their widespread use in course recommendation, the aforementioned approaches all share a critical flaw: they fail to account for students' ever-evolving tastes and requirements as they progress through courses. In other words, these techniques aren't great at extracting the user's choice in every interaction, especially if the learner is interested in a wide variety of subjects and their preferences shift over time. In this scenario, the adaptivity of these techniques are not very good at keeping up with consumers' shifting tastes. the recommendation model low since is The remainder of the paper is structured as follows: In Section 2, a concise review of course recommendation is provided. The proposed technique is then described in detail in Section 3. We conducted experiments using actual data and reported the outcomes in Section 4. Section 5 concludes the document.

  Retraction notice: Efficient Course Recommendation using Deep Transformer based Ensembled Attention Model. · 2026 · DOI

• learner [14][15] preferences has become the major research endeavour to address these difficulties, as the proliferation of online courses has made it difficult for students to choose those that best suit their needs and interests. CF algorithms [19,20] are used in conventional course recommendation systems [17,18] to uncover latent information about a user's interests. Using neural recommendation algorithms grounded in deep learning techniques [21,22], these approaches deliver solid results. The Neural Attentive Recommendation Model (NARM) is one such model that can track users' sequential actions and extract their primary goals for taking the class. In addition, the fundamental recommendation model based on attention networks is trained in tandem with the Hierarchical Reinforcement Learning (HRL) model, which reduces noisy courses through learner profile model. When consumers have signed up for a wide variety of classes, the course recommendation performances have room to grow. However, HRL fails to deliver satisfactory results since it disregards the user's stated preferences. The adaptability of the recommendation model allows for a number of methods to be used in assessing learner behaviour on e-learning systems. For instance, Chen et al.'s hybrid recommender model is one such example. It does this by employing item-based CF to unearth groups of pertinent are subsequently filtered by a sequential pattern mining algorithm in consideration of commonalities in the sequence of study. Furthermore, Wan and Niu proposed a recommendation model, wherein learning objects are simulated as intelligent object entities using the self-organization theory, and the objects in these simulations naturally interact with one another. self-organization-based things, which Despite their widespread use in course recommendation, the aforementioned approaches all share a critical flaw: they fail to account for students' ever-evolving tastes and requirements as they progress through courses. In other words, these techniques aren't great at extracting the user's choice in every interaction, especially if the learner is interested in a wide variety of subjects and their preferences shift over time. In this scenario, the adaptivity of these techniques are not very good at keeping up with consumers' shifting tastes. the recommendation model low since is The remainder of the paper is structured as follows: In Section 2, a concise review of course recommendation is provided. The proposed technique is then described in detail in Section 3. We conducted experiments using actual data and reported the outcomes in Section 4. Section 5 concludes the document.

  Retraction notice: Efficient Course Recommendation using Deep Transformer based Ensembled Attention Model. · 2026 · DOI

• Despite these empirical findings, little is known about the theoretical representation power of MaxSim and how it compares to other retrieval approaches.

  Quantifying and Expanding the Theoretical Capacity of Late-Interaction Retrieval Models · 2026

• Z. Nie and Z. Nie, “A multi-source behavioral data framework for interpretable urban tourism forecasting,” Sci. Rep., vol. 16, no. 1, p. 2257, 2025, doi: 10.1038/s41598-025-32127-2. K. Zhang, Y. Chen, and C. Li, “Discovering the tourists’ behaviors and perceptions in a tourism destination by analyzing photos’ visual content with a computer deep learning model: The case of Beijing,” Tour. Manag., vol. doi: 75, https://doi.org/10.1016/j.tourman.2019.07.002. 595–608, 2019, pp. S. Puttinaovarat, S. Chai-Arayalert, and W. Saetang, “GIS-Based Personalized Tourism Recommendation Using Association Rule Mining to Support Sustainable Tourism,” Sustainability, vol. 18, no. 6, 2026, doi: 10.3390/su18063145. I. Gede, B. Arya Budaya, G. Putra, and M. Yusadara, “A Comparative Analysis of Character and Word-Based Tokenization for Kawi-Indonesian Neural Machine Translation,” Available: 2025. http://jurnal.polibatam.ac.id/index.php/JAIC [Online]. Y. Zhang, Z. Lin, C. C. Tong, and S. W. Ho, “Enhancing patterns: tokenization cumulative logic for segmenting user-generated content in logographic languages,” J. Comput. Soc. Sci., vol. 8, no. 3, p. 80, 2025, doi: 10.1007/s42001-025-00406-7. accuracy with dynamic W. Hu, Z. Huang, J. Cai, and X. Zhao, “Dual-temporal inflow–outflow dependency modeling for short-term metro outflow prediction,” PLoS One, vol. 21, no. 4, pp. e0347131-, Apr. [Online]. Available: https://doi.org/10.1371/journal.pone.0347131 2026, R. Xie, “Deep learning-based enterprise operation forecasting and green production optimization under the BRI,” Discover Artificial Intelligence, vol. 6, no. 1, p. 82, 2025, doi: 10.1007/s44163-025-00755-2. E. Celik and S. I.

  Sequential Tourism Recommendation Using Dual-Input LSTM for Sustainable Destination Distribution · 2026 · DOI

• 103369, 2023, W. Dong, Q. Kang, G. Wang, B. Zhang, and P. Liu, “Spatiotemporal behavior pattern differentiation and preference identification of tourists from the perspective of ecotourism destination based on the tourism digital footprint data,” PLoS One, vol. 18, no. 4, pp. e0285192-, Available: https://doi.org/10.1371/journal.pone.0285192 [Online]. 2023, Apr. W. Shafqat and Y.-C. Byun, “A Context-Aware Location Recommendation System for Tourists Using Hierarchical LSTM Model,” Sustainability, vol. 12, no. 10, 2020, doi: 10.3390/su12104107. C. Zeng and C.

  Sequential Tourism Recommendation Using Dual-Input LSTM for Sustainable Destination Distribution · 2026 · DOI

• regional in Review of Economics & Finance, vol. 101, p. 104246, 2025, doi: https://doi.org/10.1016/j.iref.2025.104246. L. A. Jackson, “Community-Based Tourism: A Catalyst the United Nations Sustainable for Achieving Development Goals One and Eight,” Tourism and Hospitality, doi: 10.3390/tourhosp6010029. 2025, vol. no. 6, 1, D. Wardana and A. R. Lubis, “The Role Of The Department Of Culture and Tourism Of Rokan Hulu District In The Development Of The Aek Metertu Waterfall Tourism Object,” Mar. 2023, doi: https://doi.org/10.25299/jkp.2023.vol9(1).12082. L. A. Jackson, “Community-Based Tourism: A Catalyst for Achieving the United Nations Sustainable Development Goals One and Eight,” Tourism and Hospitality, doi: 10.3390/tourhosp6010029. 2025, vol. no. 6, 1, S. P. R. Asaithambi, R. Venkatraman, and S. Venkatraman, “A Thematic Travel Recommendation System Using an Augmented Big Data Analytical Model,” Technologies (Basel)., vol. 11, no. 1, 2023, doi: 10.3390/technologies11010028. J. Yu and R. Egger, “Tourist Experiences at Overcrowded Attractions: A Text Analytics Approach,” in Information and Communication Technologies in Tourism 2021, W. Wörndl, C. Koo, and J. L. Stienmetz, Eds., Cham: Springer International Publishing, 2021, pp. 231–243. Y. Wei and M. Chen, “Tracing the Evolution of Tourist Perception of Destination Image: A Multi-Method Analysis of a Cultural Heritage Tourist Site,” Sustainability, doi: vol. 10.3390/su17125476. 2025, no. 17, 12, H. Alshamlan, G. Alghofaili, N. ALFulayj, S. Aldawsari, Y. Alrubaiya, and R.

  Sequential Tourism Recommendation Using Dual-Input LSTM for Sustainable Destination Distribution · 2026 · DOI

Most-cited papers in Recommender Systems and Techniques

• A Review of Content-Based and Context-Based Recommendation Systems · International Journal of Emerging Technologies in Learning (iJET) · 2021 · 301 citations
• LLMRec: Large Language Models with Graph Augmentation for Recommendation · 2024 · 218 citations
• Representation Learning with Large Language Models for Recommendation · 2024 · 178 citations
• How Can Recommender Systems Benefit from Large Language Models: A Survey · ACM Transactions on Information Systems · 2024 · 161 citations
• DiffKG: Knowledge Graph Diffusion Model for Recommendation · 2024 · 135 citations
• Data-efficient Fine-tuning for LLM-based Recommendation · 2024 · 133 citations
• Evaluating Retrieval Quality in Retrieval-Augmented Generation · 2024 · 131 citations
• A context-aware citation recommendation model with BERT and graph convolutional networks · Scientometrics · 2020 · 129 citations
• LGMRec: Local and Global Graph Learning for Multimodal Recommendation · Proceedings of the AAAI Conference on Artificial Intelligence · 2024 · 116 citations
• When latent features meet side information: A preference relation based graph neural network for collaborative filtering · Expert Systems with Applications · 2024 · 116 citations

Most recent work

• A Sentiment-Aware and Explainable Hybrid Recommender System Based on Ratings and Transformer Embeddings · International Journal of Computational Intelligence Systems · 2026
• RecoWorld: Building Simulated Environments for Agentic Recommender Systems · 2026
• CooSBR: Rethinking neighborhood integration for session-based recommendation · Information Processing & Management · 2026
• Compositions of Variant Experts for Integrating Short-Term and Long-Term Preferences · ACM Transactions on Recommender Systems · 2026
• An enhanced hybrid recommender system using an adaptive optimization approach · Discover Artificial Intelligence · 2026
• Multimodal Metric Learning for Fine-Grained Video Retention Prediction in Ego-Centric Networks · Artificial Intelligence · 2026
• DeepSentRec: a deep learning-based sentiment-aware product recommendation system · Scientific Reports · 2026
• Beyond Hybrids: Superior Recommendation System via Probabilistic Similarity Measures and Model Simplification · International Journal of Intelligent Engineering and Systems · 2026
• AI-Driven Recommendation System for Personalized Insurance Solutions · Zenodo (CERN European Organization for Nuclear Research) · 2026
• Beyond Single Sequence: User Correlation Guided Cross-sequence Mixing for Sequential Recommendation · Machine Intelligence Research · 2026

Related topics in Computer Science