{"type":"ScholarlyArticle","title":"NS4S: Neighborhood Search for Scheduling Problems Via Large Language Models","authors":[{"name":"Junjie Zhang","slug":"junjie-zhang","given":"Junjie","family":"Zhang","affiliation":""},{"name":"Canhui Luo","slug":"canhui-luo","given":"Canhui","family":"Luo","affiliation":""},{"name":"Zhouxing Su","slug":"zhouxing-su","given":"Zhouxing","family":"Su","affiliation":""},{"name":"Qingyun Zhang","slug":"qingyun-zhang","given":"Qingyun","family":"Zhang","affiliation":""},{"name":"Zhipeng Lü","slug":"zhipeng-lu","given":"Zhipeng","family":"Lü","affiliation":""},{"name":"Junwen Ding","slug":"junwen-ding","given":"Junwen","family":"Ding","affiliation":""},{"name":"Yan Jin","slug":"yan-jin","given":"Yan","family":"Jin","affiliation":""}],"abstract":"Large Language Models (LLMs) have emerged as a promising technology for solving combinatorial optimization problems. However, their direct application to scheduling problems remains limited due to the inherent complexity of these problems. This paper proposes an LLMs-based neighborhood search method that leverages LLMs to tackle the job shop scheduling problem (JSP) and its variants.\r\nThe main contributions of this work are threefold. First, we introduce a novel LLMs-guided neighborhood evaluation strategy that guides local search by dynamically adjusting operation weights. Second, we develop a verification evolution (VeEvo) framework to mitigate the hallucination effects of LLMs, enabling the generation of high-quality heuristics for weight updates. Third, we integrate this framework with the weighted neighborhood evaluation strategy to effectively guide the search towards promising regions.\r\nExtensive experiments are conducted on 349 benchmark instances across three classical scheduling problems. The results demonstrate that our algorithm significantly outperforms existing state-of-the-art methods. For JSP, our algorithm reduces the average optimality gap from 10.46% to 1.35% on Taillard's instances compared to reinforced adaptive staircase curriculum learning. For flexible JSP (FJSP), it reduces the gap from 13.24% to 0.05% on Brandimarte's instances compared to deep reinforcement learning methods. Furthermore, for FJSP with sequence dependent setup time, our algorithm updates 9 upper bounds for benchmark instances.","doi":"10.24963/ijcai.2025/966","url":"https://ijcai.org/proceedings/2025/966","pdf":"https://ijcai.org/proceedings/2025/0966.pdf","markdown":"https://ijcai.org/proceedings/2025/966.md","bibtex":"https://ijcai.org/proceedings/2025/bibtex/966","volume":{"slug":"2025","title":"Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2025","url":"https://ijcai.org/proceedings/2025/","year":2025,"location":"Montreal, Canada","dates":"16-22 August 2025","editors":"James Kwok","isbn":"978-1-956792-06-5","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2025"},"section":"Planning and Scheduling","track":"Main Track","number":966,"pages":{"first":8687,"last":8695},"keywords":["Search: S: Local search","Planning and Scheduling: PS: Search in planning and scheduling","Planning and Scheduling: PS: Scheduling","Planning and Scheduling: PS: Learning in planning and scheduling"],"videos":[],"references":[],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}