{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,30]],"date-time":"2026-07-30T16:04:06Z","timestamp":1785427446870,"version":"3.56.0"},"reference-count":60,"publisher":"Association for Computing Machinery (ACM)","issue":"4","license":[{"start":{"date-parts":[[2026,7,30]],"date-time":"2026-07-30T00:00:00Z","timestamp":1785369600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/legalcode"}],"funder":[{"name":"NAVER-Intel Co-Lab"},{"name":"Institute of Information & Communications Technology Planning & Evaluation"},{"name":"Korea government"},{"name":"EntireDB2AI: Foundations and Software for Comprehensive Deep Representation Learning and Prediction on Entire Relational Databases, 40%","award":["RS-2024-00438638"],"award-info":[{"award-number":["RS-2024-00438638"]}]},{"name":"Artificial Intelligence Graduate School Program (KAIST), 10%","award":["RS-2019-II190075"],"award-info":[{"award-number":["RS-2019-II190075"]}]}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Recomm. Syst."],"published-print":{"date-parts":[[2026,12,31]]},"abstract":"<jats:p>Graph Neural Networks (GNNs) have emerged as effective tools in recommender systems. Among various GNN models, LightGCN is distinguished by its simplicity and outstanding performance. Its efficiency has led to widespread adoption across different domains, including social, bundle, and multimedia recommendations. In this article, we thoroughly examine the mechanisms of LightGCN, focusing on its strategies for scaling embeddings, aggregating neighbors, and pooling embeddings across layers. Our analysis reveals that, contrary to expectations based on its design, LightGCN suffers from inflexibility and inconsistency when applied to real-world data.<\/jats:p>\n                  <jats:p>We introduce LightGCN++, an enhanced version of LightGCN designed to address the identified limitations. LightGCN++ incorporates flexible scaling of embedding norms and neighbor weighting, along with a tailored approach for pooling layer-wise embeddings to resolve the identified inconsistencies. Despite being a remarkably simple remedy, extensive experimental results demonstrate that LightGCN++ significantly outperforms LightGCN, achieving an improvement of up to 29.38% in terms of NDCG@10. Furthermore, state-of-the-art models utilizing LightGCN as a backbone for item, bundle, multimedia, and knowledge-graph-based recommendations exhibit improved performance when equipped with LightGCN++.<\/jats:p>","DOI":"10.1145\/3760763","type":"journal-article","created":{"date-parts":[[2025,8,13]],"date-time":"2025-08-13T11:15:33Z","timestamp":1755083733000},"page":"1-26","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":3,"title":["Revisiting LightGCN: Unexpected Inflexibility, Inconsistency, and A Remedy Towards Improved Recommendation"],"prefix":"10.1145","volume":"4","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-6339-9758","authenticated-orcid":false,"given":"Geon","family":"Lee","sequence":"first","affiliation":[{"name":"KAIST","place":["Seoul, Korea (the Republic of)"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0008-8304-9585","authenticated-orcid":false,"given":"Kyungho","family":"Kim","sequence":"additional","affiliation":[{"name":"KAIST","place":["Seoul, Korea (the Republic of)"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0497-3902","authenticated-orcid":false,"given":"Fanchen","family":"Bu","sequence":"additional","affiliation":[{"name":"KAIST","place":["Daejeon, Korea (the Republic of)"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8919-0215","authenticated-orcid":false,"given":"Langzhang","family":"Liang","sequence":"additional","affiliation":[{"name":"KAIST","place":["Seoul, Korea (the Republic of)"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2872-1526","authenticated-orcid":false,"given":"Kijung","family":"Shin","sequence":"additional","affiliation":[{"name":"KAIST","place":["Seoul, Korea (the Republic of)"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2026,7,30]]},"reference":[{"key":"e_1_3_3_2_2","volume-title":"Proceedings of the ICLR","author":"Cai Xuheng","year":"2022","unstructured":"Xuheng Cai, Chao Huang, Lianghao Xia, and Xubin Ren. 2022. LightGCL: Simple yet effective graph contrastive learning for recommendation. In Proceedings of the ICLR."},{"key":"e_1_3_3_3_2","volume-title":"Proceedings of the AAAI","author":"Chen Lei","year":"2020","unstructured":"Lei Chen, Le Wu, Richang Hong, Kun Zhang, and Meng Wang. 2020. Revisiting graph based collaborative filtering: A linear residual graph convolutional network approach. In Proceedings of the AAAI."},{"key":"e_1_3_3_4_2","volume-title":"Proceedings of the ICML","author":"Chen Ting","year":"2020","unstructured":"Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. 2020. A simple framework for contrastive learning of visual representations. In Proceedings of the ICML."},{"key":"e_1_3_3_5_2","doi-asserted-by":"publisher","DOI":"10.1145\/3292500.3330652"},{"key":"e_1_3_3_6_2","volume-title":"Proceedings of the KDD","author":"Cho Eunjoon","year":"2011","unstructured":"Eunjoon Cho, Seth A Myers, and Jure Leskovec. 2011. Friendship and mobility: User movement in location-based social networks. In Proceedings of the KDD."},{"key":"e_1_3_3_7_2","doi-asserted-by":"publisher","DOI":"10.1145\/3637067"},{"key":"e_1_3_3_8_2","doi-asserted-by":"publisher","DOI":"10.1145\/3568022"},{"key":"e_1_3_3_9_2","volume-title":"Proceedings of the EMNLP","author":"Gao Tianyu","year":"2021","unstructured":"Tianyu Gao, Xingcheng Yao, and Danqi Chen. 2021. SimCSE: Simple contrastive learning of sentence embeddings. In Proceedings of the EMNLP."},{"key":"e_1_3_3_10_2","volume-title":"Proceedings of the NIPS","author":"Hamilton Will","year":"2017","unstructured":"Will Hamilton, Zhitao Ying, and Jure Leskovec. 2017. Inductive representation learning on large graphs. In Proceedings of the NIPS."},{"key":"e_1_3_3_11_2","doi-asserted-by":"publisher","DOI":"10.1145\/2827872"},{"key":"e_1_3_3_12_2","doi-asserted-by":"publisher","DOI":"10.1145\/3397271.3401063"},{"key":"e_1_3_3_13_2","doi-asserted-by":"publisher","DOI":"10.1145\/3038912.3052569"},{"key":"e_1_3_3_14_2","volume-title":"Proceedings of the RecSys","author":"Iv\u00e1n Cantador","year":"2011","unstructured":"Cantador Iv\u00e1n, Brusilovsky Peter, and Kuflik Tsvi. 2011. Second workshop on information heterogeneity and fusion in recommender systems. In Proceedings of the RecSys."},{"key":"e_1_3_3_15_2","volume-title":"Proceedings of the NeurIPS","author":"Ju Mingxuan","year":"2024","unstructured":"Mingxuan Ju, William Shiao, Zhichun Guo, Yanfang Ye, Yozen Liu, Neil Shah, and Tong Zhao. 2024. How does message passing improve collaborative filtering?. In Proceedings of the NeurIPS."},{"key":"e_1_3_3_16_2","volume-title":"Proceedings of the PAKDD","author":"Kim Kyungho","year":"2025","unstructured":"Kyungho Kim, Sunwoo Kim, Geon Lee, Jinhong Jung, and Kijung Shin. 2025. Multi-behavior recommender systems: A survey. In Proceedings of the PAKDD."},{"key":"e_1_3_3_17_2","volume-title":"Proceedings of the CIKM","author":"Kim Kyungho","year":"2024","unstructured":"Kyungho Kim, Sunwoo Kim, Geon Lee, and Kijung Shin. 2024. Towards better utilization of multiple views for bundle recommendation. In Proceedings of the CIKM."},{"key":"e_1_3_3_18_2","doi-asserted-by":"publisher","DOI":"10.1145\/3511808.3557387"},{"key":"e_1_3_3_19_2","volume-title":"Proceedings of the ICLR","author":"Kipf Thomas N","year":"2017","unstructured":"Thomas N Kipf and Max Welling. 2017. Semi-supervised classification with graph convolutional networks. In Proceedings of the ICLR."},{"key":"e_1_3_3_20_2","doi-asserted-by":"publisher","DOI":"10.1109\/MC.2009.263"},{"key":"e_1_3_3_21_2","volume-title":"Proceedings of the CIKM","author":"Lee Geon","year":"2024","unstructured":"Geon Lee, Kyungho Kim, and Kijung Shin. 2024. Post-training embedding enhancement for long-tail recommendation. In Proceedings of the CIKM."},{"key":"e_1_3_3_22_2","volume-title":"Proceedings of the RecSys","author":"Lee Geon","year":"2024","unstructured":"Geon Lee, Kyungho Kim, and Kijung Shin. 2024. Revisiting LightGCN: Unexpected inflexibility, inconsistency, and a remedy towards improved recommendation. In Proceedings of the RecSys."},{"key":"e_1_3_3_23_2","volume-title":"Proceedings of the WSDM","author":"Li Kang","year":"2023","unstructured":"Kang Li, Chang-Dong Wang, Jian-Huang Lai, and Huaqiang Yuan. 2023. Self-supervised group graph collaborative filtering for group recommendation. In Proceedings of the WSDM."},{"key":"e_1_3_3_24_2","volume-title":"Proceedings of the AAAI","author":"Li Qimai","year":"2018","unstructured":"Qimai Li, Zhichao Han, and Xiao-Ming Wu. 2018. Deeper insights into graph convolutional networks for semi-supervised learning. In Proceedings of the AAAI."},{"key":"e_1_3_3_25_2","volume-title":"Proceedings of the WWW","author":"Liang Dawen","year":"2016","unstructured":"Dawen Liang, Laurent Charlin, James McInerney, and David M Blei. 2016. Modeling user exposure in recommendation. In Proceedings of the WWW."},{"key":"e_1_3_3_26_2","volume-title":"Proceedings of the WWW","author":"Liang Dawen","year":"2018","unstructured":"Dawen Liang, Rahul G Krishnan, Matthew D Hoffman, and Tony Jebara. 2018. Variational autoencoders for collaborative filtering. In Proceedings of the WWW."},{"key":"e_1_3_3_27_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.ins.2022.01.001"},{"key":"e_1_3_3_28_2","volume-title":"Proceedings of the WWW","author":"Lin Zihan","year":"2022","unstructured":"Zihan Lin, Changxin Tian, Yupeng Hou, and Wayne Xin Zhao. 2022. Improving graph collaborative filtering with neighborhood-enriched contrastive learning. In Proceedings of the WWW."},{"key":"e_1_3_3_29_2","volume-title":"Proceedings of the WWW","author":"Liu Fan","year":"2021","unstructured":"Fan Liu, Zhiyong Cheng, Lei Zhu, Zan Gao, and Liqiang Nie. 2021. Interest-aware message-passing gcn for recommendation. In Proceedings of the WWW."},{"key":"e_1_3_3_30_2","doi-asserted-by":"publisher","DOI":"10.3233\/IDA-205725"},{"key":"e_1_3_3_31_2","volume-title":"Proceedings of the KDD","author":"Ma Yunshan","year":"2022","unstructured":"Yunshan Ma, Yingzhi He, An Zhang, Xiang Wang, and Tat-Seng Chua. 2022. CrossCBR: Cross-view contrastive learning for bundle recommendation. In Proceedings of the KDD."},{"key":"e_1_3_3_32_2","doi-asserted-by":"crossref","unstructured":"Zhongyu Ouyang Mingxuan Ju Soroush Vosoughi and Yanfang Ye. 2025. Non-parametric graph convolution for re-ranking in recommendation systems. arXiv:2507.09969. Retrieved from https:\/\/arxiv.org\/abs\/2507.09969","DOI":"10.1145\/3705328.3748058"},{"key":"e_1_3_3_33_2","volume-title":"Proceedings of the KDD","author":"Peng Shaowen","year":"2024","unstructured":"Shaowen Peng, Xin Liu, Kazunari Sugiyama, and Tsunenori Mine. 2024. How powerful is graph filtering for recommendation. In Proceedings of the KDD."},{"key":"e_1_3_3_34_2","volume-title":"Proceedings of the UAI","author":"Rendle Steffen","year":"2009","unstructured":"Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme. 2009. BPR: Bayesian personalized ranking from implicit feedback. In Proceedings of the UAI."},{"key":"e_1_3_3_35_2","volume-title":"Proceedings of the NIPS Workshop on Machine Learning for eCommerce","author":"Strub Florian","year":"2015","unstructured":"Florian Strub, Jeremie Mary, and Preux Philippe. 2015. Collaborative filtering with stacked denoising autoencoders and sparse inputs. In Proceedings of the NIPS Workshop on Machine Learning for eCommerce."},{"key":"e_1_3_3_36_2","volume-title":"Proceedings of the ICLR","author":"Veli\u010dkovi\u0107 Petar","year":"2018","unstructured":"Petar Veli\u010dkovi\u0107, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Li\u00f2, and Yoshua Bengio. 2018. Graph attention networks. In Proceedings of the ICLR."},{"key":"e_1_3_3_37_2","volume-title":"Proceedings of the KDD","author":"Wang Chenyang","year":"2022","unstructured":"Chenyang Wang, Yuanqing Yu, Weizhi Ma, Min Zhang, Chong Chen, Yiqun Liu, and Shaoping Ma. 2022. Towards representation alignment and uniformity in collaborative filtering. In Proceedings of the KDD."},{"key":"e_1_3_3_38_2","doi-asserted-by":"publisher","DOI":"10.1145\/3331184.3331267"},{"key":"e_1_3_3_39_2","volume-title":"Proceedings of the SIGIR","author":"Wei Yinwei","year":"2023","unstructured":"Yinwei Wei, Wenqi Liu, Fan Liu, Xiang Wang, Liqiang Nie, and Tat-Seng Chua. 2023. Lightgt: A light graph transformer for multimedia recommendation. In Proceedings of the SIGIR."},{"key":"e_1_3_3_40_2","volume-title":"Proceedings of the SIGIR","author":"Wei Yinwei","year":"2023","unstructured":"Yinwei Wei, Xiaohao Liu, Yunshan Ma, Xiang Wang, Liqiang Nie, and Tat-Seng Chua. 2023. Strategy-aware bundle recommender system. In Proceedings of the SIGIR."},{"key":"e_1_3_3_41_2","volume-title":"Proceedings of the MM","author":"Wei Yinwei","year":"2021","unstructured":"Yinwei Wei, Xiang Wang, Qi Li, Liqiang Nie, Yan Li, Xuanping Li, and Tat-Seng Chua. 2021. Contrastive learning for cold-start recommendation. In Proceedings of the MM."},{"key":"e_1_3_3_42_2","volume-title":"Proceedings of the ICML","author":"Wu Felix","year":"2019","unstructured":"Felix Wu, Amauri Souza, Tianyi Zhang, Christopher Fifty, Tao Yu, and Kilian Weinberger. 2019. Simplifying graph convolutional networks. In Proceedings of the ICML."},{"key":"e_1_3_3_43_2","volume-title":"Proceedings of the CIKM","author":"Wu Jiahao","year":"2022","unstructured":"Jiahao Wu, Wenqi Fan, Jingfan Chen, Shengcai Liu, Qing Li, and Ke Tang. 2022. Disentangled contrastive learning for social recommendation. In Proceedings of the CIKM."},{"key":"e_1_3_3_44_2","doi-asserted-by":"publisher","DOI":"10.1145\/3404835.3462862"},{"issue":"4","key":"e_1_3_3_45_2","first-page":"1","article-title":"On the effectiveness of sampled softmax loss for item recommendation","volume":"42","author":"Wu Jiancan","year":"2022","unstructured":"Jiancan Wu, Xiang Wang, Xingyu Gao, Jiawei Chen, Hongcheng Fu, Tianyu Qiu, and Xiangnan He. 2022. On the effectiveness of sampled softmax loss for item recommendation. ACM Transactions on Information Systems 42, 4 (2022), 1\u201326.","journal-title":"ACM Transactions on Information Systems"},{"key":"e_1_3_3_46_2","volume-title":"Proceedings of the SIGIR","author":"Wu Le","year":"2020","unstructured":"Le Wu, Yonghui Yang, Kun Zhang, Richang Hong, Yanjie Fu, and Meng Wang. 2020. Joint item recommendation and attribute inference: An adaptive graph convolutional network approach. In Proceedings of the SIGIR."},{"key":"e_1_3_3_47_2","doi-asserted-by":"publisher","DOI":"10.1145\/3535101"},{"key":"e_1_3_3_48_2","volume-title":"Proceedings of the SIGIR","author":"Xia Lianghao","year":"2022","unstructured":"Lianghao Xia, Chao Huang, Yong Xu, Jiashu Zhao, Dawei Yin, and Jimmy Huang. 2022. Hypergraph contrastive collaborative filtering. In Proceedings of the SIGIR."},{"key":"e_1_3_3_49_2","volume-title":"Proceedings of the DASFAA","author":"Xu Ronghai","year":"2023","unstructured":"Ronghai Xu, Haijun Zhao, Zhi-Yuan Li, and Chang-Dong Wang. 2023. ALGCN: Accelerated light graph convolution network for recommendation. In Proceedings of the DASFAA."},{"key":"e_1_3_3_50_2","volume-title":"Proceedings of the SIGIR","author":"Yang Yuhao","year":"2022","unstructured":"Yuhao Yang, Chao Huang, Lianghao Xia, and Chenliang Li. 2022. Knowledge graph contrastive learning for recommendation. In Proceedings of the SIGIR."},{"key":"e_1_3_3_51_2","doi-asserted-by":"publisher","DOI":"10.1145\/3219819.3219890"},{"key":"e_1_3_3_52_2","doi-asserted-by":"publisher","DOI":"10.1109\/TKDE.2023.3288135"},{"key":"e_1_3_3_53_2","volume-title":"Proceedings of the SIGIR","author":"Yu Junliang","year":"2021","unstructured":"Junliang Yu, Hongzhi Yin, Min Gao, Xin Xia, Xiangliang Zhang, and Nguyen Quoc Viet Hung. 2021. Socially-aware self-supervised tri-training for recommendation. In Proceedings of the SIGIR."},{"key":"e_1_3_3_54_2","doi-asserted-by":"publisher","DOI":"10.1109\/TKDE.2020.3033673"},{"key":"e_1_3_3_55_2","volume-title":"Proceedings of the SIGIR","author":"Yu Junliang","year":"2022","unstructured":"Junliang Yu, Hongzhi Yin, Xin Xia, Tong Chen, Lizhen Cui, and Quoc Viet Hung Nguyen. 2022. Are graph augmentations necessary? Simple graph contrastive learning for recommendation. In Proceedings of the SIGIR."},{"key":"e_1_3_3_56_2","volume-title":"Proceedings of the MM","author":"Zhang Jinghao","year":"2021","unstructured":"Jinghao Zhang, Yanqiao Zhu, Qiang Liu, Shu Wu, Shuhui Wang, and Liang Wang. 2021. Mining latent structures for multimedia recommendation. In Proceedings of the MM."},{"key":"e_1_3_3_57_2","volume-title":"Proceedings of the SIGIR","author":"Zhao Minghao","year":"2022","unstructured":"Minghao Zhao, Le Wu, Yile Liang, Lei Chen, Jian Zhang, Qilin Deng, Kai Wang, Xudong Shen, Tangjie Lv, and Runze Wu. 2022. Investigating accuracy-novelty performance for graph-based collaborative filtering. In Proceedings of the SIGIR."},{"key":"e_1_3_3_58_2","doi-asserted-by":"publisher","DOI":"10.1145\/3442381.3449788"},{"key":"e_1_3_3_59_2","volume-title":"Proceedings of the MM","author":"Zhou Xin","year":"2023","unstructured":"Xin Zhou and Zhiqi Shen. 2023. A tale of two graphs: Freezing and denoising graph structures for multimodal recommendation. In Proceedings of the MM."},{"key":"e_1_3_3_60_2","doi-asserted-by":"publisher","DOI":"10.1145\/3591469"},{"key":"e_1_3_3_61_2","volume-title":"Proceedings of the SIGIR","author":"Zou Ding","year":"2022","unstructured":"Ding Zou, Wei Wei, Xian-Ling Mao, Ziyang Wang, Minghui Qiu, Feida Zhu, and Xin Cao. 2022. Multi-level cross-view contrastive learning for knowledge-aware recommender system. In Proceedings of the SIGIR."}],"container-title":["ACM Transactions on Recommender Systems"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3760763","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,30]],"date-time":"2026-07-30T15:39:16Z","timestamp":1785425956000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3760763"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,30]]},"references-count":60,"journal-issue":{"issue":"4","published-print":{"date-parts":[[2026,12,31]]}},"alternative-id":["10.1145\/3760763"],"URL":"https:\/\/doi.org\/10.1145\/3760763","relation":{},"ISSN":["2770-6699"],"issn-type":[{"value":"2770-6699","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,30]]},"assertion":[{"value":"2025-02-15","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-07-23","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-07-30","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}