{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,10]],"date-time":"2026-04-10T15:09:57Z","timestamp":1775833797779,"version":"3.50.1"},"reference-count":36,"publisher":"Association for Computing Machinery (ACM)","issue":"4","funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"crossref","award":["62272008"],"award-info":[{"award-number":["62272008"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"crossref"}]},{"name":"ZTE-PKU joint program","award":["HC-CN-20210319008"],"award-info":[{"award-number":["HC-CN-20210319008"]}]}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Knowl. Discov. Data"],"published-print":{"date-parts":[[2026,5,31]]},"abstract":"<jats:p>\n                    Learned plan-selection optimizers combine the conventional and learned approaches by generating diverse candidate plans through multiple optimizers and selecting the best via value models. However, these eagerly-generated plans incur high optimization overhead, as they require multiple invocations of the native optimizer. In this article, we propose MoEPlan, which learns a routing policy to select top-\n                    <jats:inline-formula content-type=\"math\/tex\">\n                      <jats:tex-math notation=\"LaTeX\" version=\"MathJax\">\\( k \\)<\/jats:tex-math>\n                    <\/jats:inline-formula>\n                    experts (different optimizers) via query embedding and learnable parameters, avoiding pre-generation of candidate plans. Our approach integrates two optimization strategies: (1) a virtual ideal expert to guide the best plan selection through learned plan similarities, and (2) a query-irrelevant expert sampling strategy to balance the training cost and effectiveness of selected plans in the first round of expert selection. Furthermore, we design a two-phase training process: the first phase pre-trains the model with complete expert feedback, while the second phase filters the full expert pool to yield a promising subset and refines the selection to pinpoint the optimal expert. Experimental studies show that MoEPlan, with only two plans generated, takes less inference time, while still producing more efficient plans than other learned plan-selection optimizers.\n                  <\/jats:p>","DOI":"10.1145\/3800575","type":"journal-article","created":{"date-parts":[[2026,3,4]],"date-time":"2026-03-04T12:38:11Z","timestamp":1772627891000},"page":"1-26","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["MoEPlan: A Lazy Learned Query-Selection Optimizer via Mixture of Optimizer Experts"],"prefix":"10.1145","volume":"20","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-2580-8068","authenticated-orcid":false,"given":"Suchen","family":"Liu","sequence":"first","affiliation":[{"name":"Key Laboratory of High Confidence Software Technologies, CS, Peking University, Beijing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6750-8496","authenticated-orcid":false,"given":"Jun","family":"Gao","sequence":"additional","affiliation":[{"name":"Key Laboratory of High Confidence Software Technologies, CS, Peking University, Beijing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5578-2351","authenticated-orcid":false,"given":"Yinjun","family":"Han","sequence":"additional","affiliation":[{"name":"ZTE Corporation, Nanjing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0007-8658-0016","authenticated-orcid":false,"given":"Yang","family":"Lin","sequence":"additional","affiliation":[{"name":"ZTE Corporation, Nanjing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"320","published-online":{"date-parts":[[2026,4,10]]},"reference":[{"key":"e_1_3_1_2_2","first-page":"397","article-title":"Using confidence bounds for exploitation-exploration trade-offs","volume":"3","author":"Auer Peter","year":"2002","unstructured":"Peter Auer. 2002. Using confidence bounds for exploitation-exploration trade-offs. J. Mach. Learn. Res. 3 (2002), 397\u2013422.","journal-title":"J. Mach. Learn. Res"},{"key":"e_1_3_1_3_2","doi-asserted-by":"publisher","DOI":"10.1137\/S0097539701398375"},{"key":"e_1_3_1_4_2","doi-asserted-by":"publisher","DOI":"10.14778\/3587136.3587150"},{"key":"e_1_3_1_5_2","doi-asserted-by":"publisher","DOI":"10.5555\/3600270.3601945"},{"key":"e_1_3_1_6_2","first-page":"4171","volume-title":"Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT \u201919)","author":"Devlin Jacob","year":"2019","unstructured":"Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT \u201919), 4171\u20134186."},{"key":"e_1_3_1_7_2","unstructured":"Dujian Ding Ankur Mallick Chi Wang Robert Sim Subhabrata Mukherjee Victor Ruhle Laks V. S. Lakshmanan and Ahmed Hassan Awadallah. 2024. Hybrid LLM: Cost-efficient and quality-aware query routing. arXiv:2404.14618. Retrieved from https:\/\/arxiv.org\/abs\/2404.14618"},{"key":"e_1_3_1_8_2","unstructured":"Vijay Prakash Dwivedi and Xavier Bresson. 2020. A generalization of transformer networks to graphs. arXiv:2012.09699. Retrieved from https:\/\/arxiv.org\/abs\/arXiv:2012.09699"},{"key":"e_1_3_1_9_2","doi-asserted-by":"publisher","DOI":"10.5555\/3586589.3586709"},{"key":"e_1_3_1_10_2","volume-title":"Proceedings of the 13th International Conference on Learning Representations","author":"Feng Tao","year":"2024","unstructured":"Tao Feng, Yanzhen Shen, and Jiaxuan You. 2024. GraphRouter: A graph-based router for LLM selections. In Proceedings of the 13th International Conference on Learning Representations."},{"key":"e_1_3_1_11_2","doi-asserted-by":"publisher","DOI":"10.5555\/3104322.3104326"},{"key":"e_1_3_1_12_2","doi-asserted-by":"publisher","DOI":"10.1145\/1270.1498"},{"key":"e_1_3_1_13_2","doi-asserted-by":"publisher","DOI":"10.1162\/neco.1991.3.1.79"},{"key":"e_1_3_1_14_2","doi-asserted-by":"publisher","DOI":"10.14778\/2850583.2850594"},{"key":"e_1_3_1_15_2","doi-asserted-by":"publisher","DOI":"10.1145\/1772690.1772758"},{"key":"e_1_3_1_16_2","unstructured":"Kuan-Ming Liu and Ming-Chih Lo. 2025. LLM-Based routing in mixture of experts: A novel framework for trading. arXiv:2501.09636. Retrieved from https:\/\/arxiv.org\/abs\/2501.09636"},{"key":"e_1_3_1_17_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-20053-3_26"},{"key":"e_1_3_1_18_2","doi-asserted-by":"publisher","DOI":"10.1145\/3448016.3452838"},{"key":"e_1_3_1_19_2","unstructured":"Ryan Marcus Parimarjan Negi Hongzi Mao Chi Zhang Mohammad Alizadeh Tim Kraska Olga Papaemmanouil and Nesime Tatbul. 2019. Neo: A learned query optimizer. arXiv:1904.03711. Retrieved from https:\/\/arxiv.org\/abs\/1904.03711"},{"key":"e_1_3_1_20_2","doi-asserted-by":"publisher","DOI":"10.14778\/3342263.3342646"},{"key":"e_1_3_1_21_2","doi-asserted-by":"publisher","DOI":"10.5555\/3015812.3016002"},{"key":"e_1_3_1_22_2","unstructured":"Quang H. Nguyen Duy C. Hoang Juliette Decugis Saurav Manchanda Nitesh V. Chawla and Khoa D. Doan. 2024. MetaLLM: A high-performant and cost-efficient dynamic framework for wrapping LLMs. arXiv:2407.10834. Retrieved from https:\/\/arxiv.org\/abs\/2407.10834"},{"key":"e_1_3_1_23_2","volume-title":"Proceedings of the 13th International Conference on Learning Representations","author":"Ong Isaac","year":"2024","unstructured":"Isaac Ong, Amjad Almahairi, Vincent Wu, Wei-Lin Chiang, Tianhao Wu, Joseph E. Gonzalez, M. Waleed Kadous, and Ion Stoica. 2024. RouteLLM: Learning to route LLMs from preference data. In Proceedings of the 13th International Conference on Learning Representations."},{"key":"e_1_3_1_24_2","doi-asserted-by":"publisher","DOI":"10.1145\/564691.564759"},{"key":"e_1_3_1_25_2","unstructured":"Postgresql team. 2026. pg-hint-plan 1.6. Retrieved from https:\/\/pg-hint-plan.readthedocs.io\/en\/latest\/"},{"key":"e_1_3_1_26_2","doi-asserted-by":"publisher","DOI":"10.1561\/2200000070"},{"key":"e_1_3_1_27_2","doi-asserted-by":"crossref","unstructured":"Dimitris Stripelis Zijian Hu Jipeng Zhang Zhaozhuo Xu Alay Dilipbhai Shah Han Jin Yuhang Yao Salman Avestimehr and Chaoyang He. 2024. TensorOpera router: A multi-model router for efficient LLM inference. arXiv:2408.12320. 2024. Retrieved from https:\/\/arxiv.org\/abs\/2408.12320","DOI":"10.18653\/v1\/2024.emnlp-industry.34"},{"key":"e_1_3_1_28_2","doi-asserted-by":"publisher","DOI":"10.14778\/3368289.3368296"},{"key":"e_1_3_1_29_2","first-page":"1556","volume-title":"Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics","author":"Sheng Tai Kai","year":"2015","unstructured":"Kai Sheng Tai, Richard Socher, and Christopher D. Manning. 2015. Improved semantic representations from tree-structured long short-term memory networks. In Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics, 1556\u20131566."},{"key":"e_1_3_1_30_2","doi-asserted-by":"publisher","DOI":"10.14778\/3611479.3611528"},{"key":"e_1_3_1_31_2","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2020.2978386"},{"key":"e_1_3_1_32_2","doi-asserted-by":"publisher","DOI":"10.14778\/3611540.3611576"},{"key":"e_1_3_1_33_2","doi-asserted-by":"publisher","DOI":"10.1145\/3514221.3517885"},{"key":"e_1_3_1_34_2","doi-asserted-by":"publisher","DOI":"10.14778\/3421424.3421432"},{"key":"e_1_3_1_35_2","first-page":"1297","volume-title":"Proceedings of the 2020 IEEE 36th International Conference on Data Engineering (ICDE)","author":"Yu Xiang","year":"2020","unstructured":"Xiang Yu, Guoliang Li, Chengliang Chai, and Nan Tang. 2020. Reinforcement learning with Tree-LSTM for join order selection. In Proceedings of the 2020 IEEE 36th International Conference on Data Engineering (ICDE). IEEE, 1297\u20131308."},{"key":"e_1_3_1_36_2","doi-asserted-by":"publisher","DOI":"10.14778\/3529337.3529349"},{"key":"e_1_3_1_37_2","doi-asserted-by":"publisher","DOI":"10.14778\/3583140.3583160"}],"container-title":["ACM Transactions on Knowledge Discovery from Data"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3800575","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,4,10]],"date-time":"2026-04-10T14:41:18Z","timestamp":1775832078000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3800575"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,10]]},"references-count":36,"journal-issue":{"issue":"4","published-print":{"date-parts":[[2026,5,31]]}},"alternative-id":["10.1145\/3800575"],"URL":"https:\/\/doi.org\/10.1145\/3800575","relation":{},"ISSN":["1556-4681","1556-472X"],"issn-type":[{"value":"1556-4681","type":"print"},{"value":"1556-472X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,4,10]]},"assertion":[{"value":"2025-03-23","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-02-18","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-04-10","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}