{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,26]],"date-time":"2026-06-26T18:45:33Z","timestamp":1782499533573,"version":"3.54.5"},"reference-count":27,"publisher":"Public Library of Science (PLoS)","issue":"6","license":[{"start":{"date-parts":[[2026,6,26]],"date-time":"2026-06-26T00:00:00Z","timestamp":1782432000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Van Lang University, Ho Chi Minh City, Vietnam"},{"name":"Nguyen Tat Thanh University, Ho Chi Minh City, Vietnam"}],"content-domain":{"domain":["www.plosone.org"],"crossmark-restriction":false},"short-container-title":["PLoS One"],"abstract":"<jats:p>In recent years, graph neural networks (GNNs) have emerged as a dominant paradigm for learning from graph-structured data, achieving strong performance across a wide range of graph mining tasks, including node classification, clustering, and relational inference. Despite these advances, most existing GNN architectures primarily rely on message-passing mechanisms that aggregate smoothed high-order local neighborhood information. While effective for capturing local proximity patterns, such designs often struggle to model richer multi-level graph structures, particularly in scenarios where both fine-grained local interactions and holistic global organization are essential. From a structural perspective, graph representation learning inherently involves information distributed across multiple levels. Local proximity cues, such as node degrees and common neighbors, capture immediate relational patterns, whereas global structural characteristics, including long-range dependencies and overall graph topology, provide complementary high-order information. However, conventional GNN models typically entangle these heterogeneous signals within a unified aggregation process, leading to suboptimal representations. To address this limitation, we propose DAGRL, a dual attention-based graph representation learning framework that explicitly disentangles and integrates multi-level structural information. The proposed model employs dedicated graph neural components to learn structure-specific representations and utilizes a dual attention mechanism to adaptively fuse them into a unified embedding space. Extensive experiments demonstrate that DAGRL consistently outperforms existing methods across multiple benchmark datasets.<\/jats:p>","DOI":"10.1371\/journal.pone.0351922","type":"journal-article","created":{"date-parts":[[2026,6,26]],"date-time":"2026-06-26T17:46:01Z","timestamp":1782495961000},"page":"e0351922","update-policy":"https:\/\/doi.org\/10.1371\/journal.pone.corrections_policy","source":"Crossref","is-referenced-by-count":0,"title":["An approach of dual attention-driven multi-levelled graph neural architecture for rich-structural social graph analysis and representation learning"],"prefix":"10.1371","volume":"21","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-3990-312X","authenticated-orcid":true,"given":"Linh Nguyen Thi","family":"My","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7291-4168","authenticated-orcid":true,"given":"Tham","family":"Vo","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,6,26]]},"reference":[{"issue":"1","key":"pone.0351922.ref001","article-title":"Link prediction based on spectral analysis","volume":"19","author":"C Gui","year":"2024","journal-title":"PLoS One"},{"key":"pone.0351922.ref002","doi-asserted-by":"crossref","first-page":"124289","DOI":"10.1016\/j.physa.2020.124289","article-title":"Link prediction techniques, applications, and performance: a survey","volume":"553","author":"A Kumar","year":"2020","journal-title":"Physica A: Stat Mech Appl"},{"issue":"1","key":"pone.0351922.ref003","doi-asserted-by":"crossref","first-page":"4","DOI":"10.1109\/TNNLS.2020.2978386","article-title":"A comprehensive survey on graph neural networks","volume":"32","author":"Z Wu","year":"2021","journal-title":"IEEE Trans Neural Netw Learn Syst"},{"issue":"6","key":"pone.0351922.ref004","doi-asserted-by":"crossref","DOI":"10.1371\/journal.pone.0304858","article-title":"BERT4FCA: A method for bipartite link prediction using formal concept analysis and BERT","volume":"19","author":"S Peng","year":"2024","journal-title":"PLoS One"},{"issue":"9","key":"pone.0351922.ref005","doi-asserted-by":"crossref","DOI":"10.1371\/journal.pone.0308544","article-title":"Evolution mechanism of industrial network in Yangtze River Delta region from the perspective of link prediction","volume":"19","author":"Y Shen","year":"2024","journal-title":"PLoS One"},{"key":"pone.0351922.ref006","first-page":"111","article-title":"Impact of Routing Techniques and Mobility Models on Flying Ad Hoc Networks","author":"MA Hassan","year":"2022","journal-title":"Studies in Computational Intelligence for Unmanned Aerial Vehicles Communication Networks"},{"key":"pone.0351922.ref007","article-title":"Harnessing 1D-CNN for received power prediction in sub-6 GHz RIS: part I","author":"MA Hassan","year":"2025","journal-title":"2025 IEEE International Conference on Communications Workshops"},{"issue":"5","key":"pone.0351922.ref008","doi-asserted-by":"crossref","first-page":"833","DOI":"10.1109\/TKDE.2018.2849727","article-title":"A survey on network embedding","volume":"31","author":"P Cui","year":"2019","journal-title":"IEEE Trans Knowl Data Eng"},{"key":"pone.0351922.ref009","first-page":"701","article-title":"Deepwalk: Online learning of social representations.","author":"B Perozzi","year":"2014","journal-title":"SIGKDD"},{"key":"pone.0351922.ref010","doi-asserted-by":"crossref","first-page":"1067","DOI":"10.1145\/2736277.2741093","article-title":"Large-scale information network embedding.","author":"J Tang","year":"2015","journal-title":"WWW"},{"key":"pone.0351922.ref011","first-page":"855","article-title":"Node2vec: Scalable feature learning for networks.","author":"A Grover","year":"2016","journal-title":"SIGKDD"},{"key":"pone.0351922.ref012","article-title":"Semi-supervised classification with graph convolutional networks.","author":"TN Kipf","year":"2016","journal-title":"ICLR"},{"key":"pone.0351922.ref013","first-page":"1025","article-title":"Inductive representation learning on large graphs.","author":"W Hamilton","year":"2017","journal-title":"NeurIPS"},{"key":"pone.0351922.ref014","article-title":"Predict then propagate: Graph neural networks meet personalized PageRank.","author":"J Gasteiger","year":"2018","journal-title":"ICLR"},{"key":"pone.0351922.ref015","article-title":"Graph Attention Networks.","author":"P Veli\u010dkovi\u0107","year":"2018","journal-title":"ICLR"},{"key":"pone.0351922.ref016","article-title":"How powerful are graph neural networks?","author":"K Xu","year":"2019","journal-title":"ICLR"},{"key":"pone.0351922.ref017","first-page":"22092","article-title":"Graph random neural networks for semi-supervised learning on graphs.","author":"W Feng","year":"2020","journal-title":"NeurIPS"},{"key":"pone.0351922.ref018","first-page":"338","article-title":"Towards deeper graph neural networks.","author":"M Liu","year":"2020","journal-title":"SIGKDD"},{"key":"pone.0351922.ref019","article-title":"How Attentive are Graph Attention Networks?","author":"S Brody","year":"2022","journal-title":"ICLR"},{"key":"pone.0351922.ref020","article-title":"Variational graph auto-encoders","author":"TN Kipf","year":"2016","journal-title":"arXiv"},{"key":"pone.0351922.ref021","first-page":"5171","article-title":"Link prediction based on graph neural networks.","author":"M Zhang","year":"2018","journal-title":"NeurIPS"},{"key":"pone.0351922.ref022","first-page":"4465","article-title":"Distance encoding: Design provably more powerful neural networks for graph representation learning.","author":"P Li","year":"2020","journal-title":"NeurIPS"},{"key":"pone.0351922.ref023","article-title":"Understanding over-squashing and bottlenecks on graphs via curvature.","author":"J Topping","year":"2022","journal-title":"ICLR"},{"key":"pone.0351922.ref024","first-page":"6000","article-title":"Attention is all you need","author":"A Vaswani","year":"2017","journal-title":"NeurIPS"},{"key":"pone.0351922.ref025","first-page":"249","article-title":"Understanding the difficulty of training deep feedforward neural networks.","author":"X Glorot","year":"2010","journal-title":"AISTATS"},{"key":"pone.0351922.ref026","first-page":"4292","article-title":"The network data repository with interactive graph analytics and visualization.","author":"R Rossi","year":"2015","journal-title":"AAAI"},{"key":"pone.0351922.ref027","doi-asserted-by":"crossref","first-page":"36","DOI":"10.1145\/1134271.1134277","article-title":"The political blogosphere and the 2004 US election: divided they blog.","author":"LA Adamic","year":"2005","journal-title":"Proceedings of the 3rd International Workshop on Link discovery"}],"container-title":["PLOS One"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0351922","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,26]],"date-time":"2026-06-26T17:46:05Z","timestamp":1782495965000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0351922"}},"subtitle":[],"editor":[{"given":"Xuyun","family":"Wen","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,6,26]]},"references-count":27,"journal-issue":{"issue":"6","published-online":{"date-parts":[[2026,6,26]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pone.0351922","relation":{},"ISSN":["1932-6203"],"issn-type":[{"value":"1932-6203","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,6,26]]}}}