{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,3]],"date-time":"2026-06-03T15:28:13Z","timestamp":1780500493677,"version":"3.54.1"},"reference-count":83,"publisher":"Association for Computing Machinery (ACM)","issue":"2","funder":[{"DOI":"10.13039\/501100003977","name":"Israel Science Foundation","doi-asserted-by":"crossref","award":["2243\/20"],"award-info":[{"award-number":["2243\/20"]}],"id":[{"id":"10.13039\/501100003977","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Recomm. Syst."],"published-print":{"date-parts":[[2026,6,30]]},"abstract":"<jats:p>\n                    Recommender systems have become integral to many online services, leveraging user data to provide personalized recommendations. However, as these systems grow in complexity, understanding the rationale behind their recommendations becomes increasingly difficult. Explainable Artificial Intelligence (XAI) has emerged as a crucial field addressing this challenge, particularly in ensuring transparency and trustworthiness in automated decision-making processes. In this article, we introduce\n                    <jats:bold>Learning to eXplain Recommendations (LXR)<\/jats:bold>\n                    , a scalable, model-agnostic framework designed to generate counterfactually correct explanations for recommender systems.\n                  <\/jats:p>\n                  <jats:p>LXR generates explanations for recommendations produced by any differentiable recommender system. By leveraging both factual and counterfactual loss terms, LXR offers robust, accurate, and computationally efficient explanations that reflect the model\u2019s internal decision-making process.<\/jats:p>\n                  <jats:p>A key feature of LXR is its focus on the factual correctness of explanations through counterfactual reasoning, bridging the gap between plausible and accurate explanations. Unlike traditional approaches that rely on exhaustive perturbations of user data, LXR uses a self-supervised learning method to generate explanations efficiently, without sacrificing accuracy. LXR operates in two stages: a pre-training step and a novel Inference-Time Fine-tuning (ITF) step that refines explanations at the individual recommendation level, significantly improving accuracy with minimal computational overhead. Additionally, LXR is applied to hybrid recommender models incorporating demographic data, demonstrating its versatility across real-world scenarios. Finally, we also showcase LXR\u2019s ability to explain recommendations at various ranks within a user\u2019s recommendation list.<\/jats:p>\n                  <jats:p>\n                    As a secondary contribution, we introduce several novel evaluation metrics, inspired by saliency maps from computer vision, to rigorously assess the counterfactual correctness of explanations in recommender systems. Our results demonstrate that LXR sets a new benchmark for explainability, providing accurate, transparent, and interpretable explanations. The code is available on our GitHub repository:\n                    <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" xlink:href=\"https:\/\/github.com\/DeltaLabTLV\/LXR_\">https:\/\/github.com\/DeltaLabTLV\/LXR_<\/jats:ext-link>\n                    .\n                  <\/jats:p>","DOI":"10.1145\/3732292","type":"journal-article","created":{"date-parts":[[2025,4,24]],"date-time":"2025-04-24T06:54:41Z","timestamp":1745477681000},"page":"1-39","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":2,"title":["LXR: Learning to eXplain Recommendations"],"prefix":"10.1145","volume":"4","author":[{"ORCID":"https:\/\/orcid.org\/0009-0006-5985-5494","authenticated-orcid":false,"given":"Liya","family":"Gurevitch","sequence":"first","affiliation":[{"name":"Department of Industrial Engineering, Tel Aviv University","place":["Tel Aviv, Israel"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8005-7618","authenticated-orcid":false,"given":"Veronika","family":"Bogina","sequence":"additional","affiliation":[{"name":"Department of Industrial Engineering, Tel Aviv University","place":["Tel Aviv, Israel"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5059-0905","authenticated-orcid":false,"given":"Oren","family":"Barkan","sequence":"additional","affiliation":[{"name":"The Open University of Israel","place":["Raanana, Israel"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0000-6132-252X","authenticated-orcid":false,"given":"Yahlly","family":"Schein","sequence":"additional","affiliation":[{"name":"Department of Industrial Engineering, Tel Aviv University","place":["Tel Aviv, Israel"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0007-2816-9472","authenticated-orcid":false,"given":"Yehonatan","family":"Elisha","sequence":"additional","affiliation":[{"name":"Department of Industrial Engineering, Tel Aviv University","place":["Tel Aviv, Israel"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8219-4512","authenticated-orcid":false,"given":"Noam","family":"Koenigstein","sequence":"additional","affiliation":[{"name":"Department of Industrial Engineering, Tel Aviv University","place":["Tel Aviv, Israel"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2025,11,22]]},"reference":[{"key":"e_1_3_1_2_2","doi-asserted-by":"publisher","DOI":"10.1145\/2872518.2889405"},{"key":"e_1_3_1_3_2","doi-asserted-by":"publisher","DOI":"10.1145\/3109859.3109913"},{"key":"e_1_3_1_4_2","unstructured":"Julius Adebayo Justin Gilmer Michael Muelly Ian Goodfellow Moritz Hardt and Been Kim. 2018. Sanity checks for saliency maps. In Proceedings of the 32nd International Conference on Neural Information Processing Systems (NIPS\u201918) Montr\u00e9al Canada 9525\u20139536."},{"key":"e_1_3_1_5_2","unstructured":"Chirag Agarwal Satyapriya Krishna Eshika Saxena Martin Pawelczyk Nari Johnson Isha Puri Marinka Zitnik and Himabindu Lakkaraju. 2022. Openxai: Towards a transparent evaluation of model explanations. In Advances in Neural Information Processing Systems. 15784\u201315799."},{"key":"e_1_3_1_6_2","doi-asserted-by":"crossref","unstructured":"Sujoy Bag Sri Krishna Kumar and Manoj Kumar Tiwari. 2019. An efficient recommendation generation using relevant Jaccard similarity. Information Sciences 483 (2019) 53\u201364.","DOI":"10.1016\/j.ins.2019.01.023"},{"key":"e_1_3_1_7_2","doi-asserted-by":"publisher","unstructured":"Dor Bank Noam Koenigstein and Raja Giryes. 2023. Autoencoders. In Machine Learning for Data Science Handbook: Data Mining and Knowledge Discovery Handbook. Springer International Publishing 353\u2013374. DOI:10.1007\/978-3-031-24628-9_16","DOI":"10.1007\/978-3-031-24628-9_16"},{"key":"e_1_3_1_8_2","doi-asserted-by":"crossref","unstructured":"Oren Barkan Yuval Asher Amit Eshel Yehonatan Elisha and Noam Koenigstein. 2023. Learning to explain: A model-agnostic framework for explaining black box models. In 2023 IEEE International Conference on Data Mining (ICDM) IEEE 944\u2013949.","DOI":"10.1109\/ICDM58522.2023.00105"},{"key":"e_1_3_1_9_2","doi-asserted-by":"crossref","unstructured":"Oren Barkan Yuval Asher Amit Eshel and Noam Koenigstein. 2023. Visual explanations via iterated integrated attributions. In Proceedings of the IEEE\/CVF International Conference on Computer Vision (2023) 2073\u20132084.","DOI":"10.1109\/ICCV51070.2023.00198"},{"key":"e_1_3_1_10_2","doi-asserted-by":"publisher","DOI":"10.1145\/3589334.3645560"},{"key":"e_1_3_1_11_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP40776.2020.9053071"},{"key":"e_1_3_1_12_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICDM50108.2020.00101"},{"key":"e_1_3_1_13_2","doi-asserted-by":"publisher","DOI":"10.1145\/3383313.3412226"},{"key":"e_1_3_1_14_2","doi-asserted-by":"publisher","DOI":"10.1145\/3459637.3482126"},{"key":"e_1_3_1_15_2","doi-asserted-by":"publisher","DOI":"10.1145\/3459637.3482056"},{"key":"e_1_3_1_16_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP40776.2020.9053105"},{"key":"e_1_3_1_17_2","doi-asserted-by":"publisher","DOI":"10.1109\/MLSP.2016.7738886"},{"key":"e_1_3_1_18_2","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.acl-main.356"},{"key":"e_1_3_1_19_2","doi-asserted-by":"crossref","unstructured":"Oren Barkan Tom Shaked Yonatan Fuchs and Noam Koenigstein. 2024. Modeling users\u2019 heterogeneous taste with diversified attentive user profiles. User Modeling and User-Adapted Interaction 34 2 (2024) 375\u2013405.","DOI":"10.1007\/s11257-023-09376-9"},{"key":"e_1_3_1_20_2","doi-asserted-by":"publisher","DOI":"10.1145\/3018661.3018718"},{"key":"e_1_3_1_21_2","first-page":"153","volume-title":"Proceedings of the Beyond Personalization Workshop, IUI","volume":"5","author":"Bilgic Mustafa","year":"2005","unstructured":"Mustafa Bilgic and Raymond J Mooney. 2005. Explaining recommendations: Satisfaction vs. promotion. In Proceedings of the Beyond Personalization Workshop, IUI, Vol. 5. 153."},{"key":"e_1_3_1_22_2","doi-asserted-by":"crossref","unstructured":"L\u00e9o Brunot Nicolas Canovas Alexandre Chanson Nicolas Labroche and Will\u00e8me Verdeaux. 2022. Preference-based and local post-hoc explanations for recommender systems. Information Systems 108 (2022) 102021.","DOI":"10.1016\/j.is.2022.102021"},{"key":"e_1_3_1_23_2","doi-asserted-by":"publisher","DOI":"10.1109\/WACV.2018.00097"},{"key":"e_1_3_1_24_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR46437.2021.00084"},{"key":"e_1_3_1_25_2","doi-asserted-by":"publisher","DOI":"10.1145\/3397271.3401042"},{"key":"e_1_3_1_26_2","doi-asserted-by":"publisher","DOI":"10.1145\/3331184.3331254"},{"key":"e_1_3_1_27_2","doi-asserted-by":"publisher","DOI":"10.1145\/2911451.2911549"},{"key":"e_1_3_1_28_2","doi-asserted-by":"publisher","DOI":"10.1145\/3292500.3330857"},{"key":"e_1_3_1_29_2","first-page":"3","volume-title":"Proceedings of the KDD Cup 2011","author":"Dror Gideon","year":"2012","unstructured":"Gideon Dror, Noam Koenigstein, Yehuda Koren, and Markus Weimer. 2012. The yahoo! music dataset and kdd-cup\u201911. In Proceedings of the KDD Cup 2011. PMLR, 3\u201318."},{"key":"e_1_3_1_30_2","doi-asserted-by":"publisher","DOI":"10.1145\/3673243"},{"key":"e_1_3_1_31_2","doi-asserted-by":"crossref","unstructured":"Keren Gaiger Oren Barkan Shir Tsipory-Samuel and Noam Koenigstein. 2023. Not All memories created equal: Dynamic user representations for collaborative filtering. IEEE Access 11 (2023) 34746\u201334763.","DOI":"10.1109\/ACCESS.2023.3263931"},{"key":"e_1_3_1_32_2","doi-asserted-by":"publisher","DOI":"10.1109\/DSAA.2018.00018"},{"key":"e_1_3_1_33_2","doi-asserted-by":"publisher","DOI":"10.1609\/aimag.v38i3.2741"},{"key":"e_1_3_1_34_2","doi-asserted-by":"publisher","DOI":"10.1145\/2827872"},{"key":"e_1_3_1_35_2","doi-asserted-by":"publisher","DOI":"10.1145\/2806416.2806504"},{"key":"e_1_3_1_36_2","doi-asserted-by":"publisher","DOI":"10.1145\/3397271.3401063"},{"key":"e_1_3_1_37_2","doi-asserted-by":"publisher","DOI":"10.1145\/3038912.3052569"},{"key":"e_1_3_1_38_2","doi-asserted-by":"publisher","DOI":"10.1145\/358916.358995"},{"key":"e_1_3_1_39_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-04083-2_2"},{"key":"e_1_3_1_40_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICDM.2008.22"},{"key":"e_1_3_1_41_2","unstructured":"Yidan Hu Yong Liu Chunyan Miao Gongqi Lin and Yuan Miao. 2023. Aspect-guided syntax graph learning for explainable recommendation. IEEE Transactions on Knowledge and Data Engineering 35 8 (2023) 7768\u20137781."},{"key":"e_1_3_1_42_2","doi-asserted-by":"crossref","unstructured":"Alon Jacovi and Yoav Goldberg. 2020. Towards faithfully interpretable NLP systems: How should we define and evaluate faithfulness? In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Association for Computational Linguistics 4198\u20134205.","DOI":"10.18653\/v1\/2020.acl-main.386"},{"key":"e_1_3_1_43_2","first-page":"1885","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Koh Pang Wei","year":"2017","unstructured":"Pang Wei Koh and Percy Liang. 2017. Understanding black-box predictions via influence functions. In Proceedings of the International Conference on Machine Learning. PMLR, 1885\u20131894."},{"key":"e_1_3_1_44_2","doi-asserted-by":"publisher","DOI":"10.1109\/MC.2009.263"},{"key":"e_1_3_1_45_2","unstructured":"Matthew L. Leavitt and Ari S. Morcos. 2020. Towards falsifiable interpretability research. CoRR abs\/2010.12016 (2020). Retrieved from https:\/\/arxiv.org\/abs\/2010.12016"},{"key":"e_1_3_1_46_2","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.acl-long.383"},{"key":"e_1_3_1_47_2","doi-asserted-by":"publisher","DOI":"10.1145\/3178876.3186150"},{"key":"e_1_3_1_48_2","doi-asserted-by":"publisher","unstructured":"Scott Lundberg and Su-In Lee. 2017. A Unified approach to interpreting model predictions. In Advances in Neural Information Processing Systems. DOI:10.48550\/arXiv.1705.07874","DOI":"10.48550\/arXiv.1705.07874"},{"key":"e_1_3_1_49_2","doi-asserted-by":"publisher","DOI":"10.1145\/3523227.3546756"},{"key":"e_1_3_1_50_2","doi-asserted-by":"publisher","DOI":"10.1145\/2856767.2856813"},{"key":"e_1_3_1_51_2","doi-asserted-by":"publisher","DOI":"10.1145\/3583558"},{"key":"e_1_3_1_52_2","doi-asserted-by":"publisher","DOI":"10.1145\/3297280.3297443"},{"key":"e_1_3_1_53_2","doi-asserted-by":"publisher","DOI":"10.1145\/3219819.3220072"},{"key":"e_1_3_1_54_2","first-page":"151","volume-title":"Proceedings of the British Machine Vision Conference 2018, BMVC 2018, Newcastle, UK, September 3-6, 2018","author":"Petsiuk Vitali","year":"2018","unstructured":"Vitali Petsiuk, Abir Das, and Kate Saenko. 2018. RISE: Randomized input sampling for explanation of black-box models. In Proceedings of the British Machine Vision Conference 2018, BMVC 2018, Newcastle, UK, September 3-6, 2018. BMVA Press, 151. Retrieved from http:\/\/bmvc2018.org\/contents\/papers\/1064.pdf"},{"key":"e_1_3_1_55_2","doi-asserted-by":"publisher","DOI":"10.1162\/coli_a_00486"},{"key":"e_1_3_1_56_2","doi-asserted-by":"crossref","unstructured":"Steffen Rendle Walid Krichene Li Zhang and Yehuda Koren. 2022. Revisiting the performance of ials on item recommendation benchmarks. In Proceedings of the 16th ACM Conference on Recommender Systems. 427\u2013435.","DOI":"10.1145\/3523227.3548486"},{"key":"e_1_3_1_57_2","doi-asserted-by":"publisher","DOI":"10.1145\/3383313.3412488"},{"key":"e_1_3_1_58_2","doi-asserted-by":"publisher","DOI":"10.1145\/2939672.2939778"},{"key":"e_1_3_1_59_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-0-387-85820-3_1"},{"key":"e_1_3_1_60_2","doi-asserted-by":"publisher","unstructured":"Francesco Ricci Lior Rokach and Bracha Shapira. 2022. Recommender systems: techniques applications and challenges. In Recommender Systems Handbook. Springer US New York NY 1\u201335. DOI:10.1007\/978-1-0716-2197-4_1","DOI":"10.1007\/978-1-0716-2197-4_1"},{"key":"e_1_3_1_61_2","unstructured":"Tim Rocktaschel Edward Grefenstette Karl Moritz Hermann Tomas Kocisky and Phil Blunsom. 2016. Reasoning about Entailment with Neural Attention. In International Conference on Learning Representations (ICLR)."},{"key":"e_1_3_1_62_2","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2016.2599820"},{"key":"e_1_3_1_63_2","doi-asserted-by":"publisher","DOI":"10.1145\/3336191.3371831"},{"key":"e_1_3_1_64_2","doi-asserted-by":"publisher","DOI":"10.35940\/ijeat.E9666.069520"},{"key":"e_1_3_1_65_2","first-page":"1","volume-title":"Proceedings of the ACM SIGIR 2001 Workshop on Recommender Systems","volume":"13","author":"Swearingen Kirsten","year":"2001","unstructured":"Kirsten Swearingen and Rashmi Sinha. 2001. Beyond algorithms: An HCI perspective on recommender systems. In Proceedings of the ACM SIGIR 2001 Workshop on Recommender Systems, Vol. 13. 1\u201311."},{"key":"e_1_3_1_66_2","doi-asserted-by":"publisher","DOI":"10.1145\/1639714.1639777"},{"key":"e_1_3_1_67_2","doi-asserted-by":"publisher","DOI":"10.1145\/3459637.3482420"},{"key":"e_1_3_1_68_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICDEW.2007.4401070"},{"key":"e_1_3_1_69_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-1-4899-7637-6_10"},{"key":"e_1_3_1_70_2","doi-asserted-by":"crossref","unstructured":"Nava Tintarev and Judith Masthoff. 2022. Beyond explaining single item recommendations. In Recommender Systems Handbook Francesco Ricci Lior Rokach and Bracha Shapira (Eds.). Springer US New York NY 711\u2013756.","DOI":"10.1007\/978-1-0716-2197-4_19"},{"key":"e_1_3_1_71_2","doi-asserted-by":"publisher","DOI":"10.1145\/3404835.3463005"},{"key":"e_1_3_1_72_2","doi-asserted-by":"publisher","DOI":"10.1145\/3331184.3331377"},{"key":"e_1_3_1_73_2","doi-asserted-by":"publisher","DOI":"10.1145\/1502650.1502661"},{"key":"e_1_3_1_74_2","doi-asserted-by":"publisher","DOI":"10.1145\/3209978.3210010"},{"key":"e_1_3_1_75_2","doi-asserted-by":"publisher","DOI":"10.1145\/3485447.3512168"},{"key":"e_1_3_1_76_2","doi-asserted-by":"publisher","DOI":"10.1145\/3331184.3331267"},{"key":"e_1_3_1_77_2","first-page":"25","volume-title":"Proceedings of the Conference on Learning Theory","author":"Wang Yining","year":"2013","unstructured":"Yining Wang, Liwei Wang, Yuanzhi Li, Di He, and Tie-Yan Liu. 2013. A theoretical analysis of NDCG type ranking measures. In Proceedings of the Conference on Learning Theory. PMLR, 25\u201354."},{"key":"e_1_3_1_78_2","doi-asserted-by":"publisher","DOI":"10.1145\/3535101"},{"key":"e_1_3_1_79_2","doi-asserted-by":"crossref","unstructured":"Yikun Xian Tong Zhao Jin Li Jim Chan Andrey Kan Jun Ma Xin Luna Dong Christos Faloutsos George Karypis S. Muthukrishnan and Yongfeng Zhang. 2021. Ex3: Explainable attribute-aware item-set recommendations. In Proceedings of the 15th ACM Conference on Recommender Systems. 484\u2013494.","DOI":"10.1145\/3460231.3474240"},{"key":"e_1_3_1_80_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-32236-6_51"},{"key":"e_1_3_1_81_2","doi-asserted-by":"publisher","DOI":"10.1145\/3219819.3219890"},{"key":"e_1_3_1_82_2","doi-asserted-by":"crossref","unstructured":"Yongfeng Zhang and Xu Chen. 2020. Explainable recommendation: A survey and new perspectives. Foundations and Trends\u00ae in Information Retrieval 14 1 (2020) 1\u2013101.","DOI":"10.1561\/1500000066"},{"key":"e_1_3_1_83_2","doi-asserted-by":"publisher","DOI":"10.1145\/2600428.2609579"},{"key":"e_1_3_1_84_2","doi-asserted-by":"publisher","DOI":"10.1145\/3477314.3507029"}],"container-title":["ACM Transactions on Recommender Systems"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3732292","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,11,22]],"date-time":"2025-11-22T16:51:11Z","timestamp":1763830271000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3732292"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,11,22]]},"references-count":83,"journal-issue":{"issue":"2","published-print":{"date-parts":[[2026,6,30]]}},"alternative-id":["10.1145\/3732292"],"URL":"https:\/\/doi.org\/10.1145\/3732292","relation":{},"ISSN":["2770-6699"],"issn-type":[{"value":"2770-6699","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,11,22]]},"assertion":[{"value":"2024-09-22","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-04-10","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-11-22","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}