计算机科学 ›› 2026, Vol. 53 ›› Issue (1): 29-38.doi: 10.11896/jsjkx.250100001
邵欣怡, 朱经纬, 张亮
SHAO Xinyi, ZHU Jingwei, ZHANG Liang
摘要： 业务流程应变是业务流程管理的重要任务,旨在通过调整流程模型和实例行为来响应不断变化的环境,从而提高其柔韧性并实现业务目标。建模时,残留不确定性导致的长尾变化无法避免,给传统的业务流程应变技术带来了挑战。目前针对长尾变化最有效的应变方法基于一种三方协作框架,即由负责感知长尾变化和提出应变策略的前端业务人员、负责提供服务接口和合规性要求的后端技术人员和管理层,以及辅助应变实施的工具系统共同协作来应对长尾变化,保障业务目标达成。然而,长尾变化在不同时空条件下的多样性、复杂性和应变的迫切性,极有可能超出前端业务人员在应变时对当前情境的理解能力、依据情境制定应变策略的专业水平,以及将应变策略采用领域专用语言有效表达的熟练程度。为弥补这一缺憾并进一步拓展上述框架,提出了一种基于大语言模型的业务流程长尾变化应变方法LLM-Adapt,充分利用大语言模型的泛化能力、强大的内容生成能力,以及嵌入的事件与对策知识库,形成一种更高效、灵活的应变机制。首先,以基于长尾变化特征的提示词工程为媒介,使前端业务人员能够通过自然语言与大语言模型进行交互并获得应变方案。其次,结合后端管理层制定的业务基线目标约束对应变方案进行功能性约束验证,提出的SSDT-Lane算法基于流程结构相似性对应变方案进行筛选,消除了大语言模型在流程调整、业务和组织架构匹配等方面面临的幻觉风险。基于合成数据和真实开源数据集的典型案例分析实验显示,LLM-Adapt相比现有方法,在应变准确性、效率、适用性等方面都表现出显著优势。
中图分类号:
| [1]WESKE M.Business Process Management—Concepts,Languages,Architectures[M].Berlin:Springer,2007. [2]CHEN X,CAO H,YE L,et al.Fulfilling functional demands of bpm in long-tailed change environments[C]//2020 International Conference on Service Science(ICSS).IEEE,2020:129-135. [3]CAO H,CHEN X,ZHANG L,et al.Guaranteeing sound reac-tions to long-tailed changes:A syntax-directed annotation approach[C]//2020 International Conference on Service Science(ICSS).IEEE,2020:162-169. [4]CAO H,CHEN X,ZHANG L,et al.A sound response to long-tailed changes in business process management[J].Service Oriented Computing and Applications,2022,16(4):279-291. [5]ZHU J,PENG J,ZHANG L,et al.Improving business processresilience to long-tailed business events via low-code[C]//2022 IEEE International Conference on Web Services(ICWS).IEEE,2022:343-348. [6]ZHU J W.Research on the Reaction to Long-tailed Changes in Business Processes[D].Shanghai:Fudan University,2023. [7]ZELLNER G.A structured evaluation of business process improvement approaches[J].Business Process Management Journal,2011,17(2):203-237. [8]KOURANI H,BERTI A,SCHUSTER D,et al.Process Mode-ling With Large Language Models[C]//International Conference on Business Process Modeling,Development and Support.Cham:Springer,2024:229-244. [9]ZHU J.On Balancing Flexibility and Compliance of BusinessProcesses:Functional Constraints Modeling and Verification[C]//International Conference on Service- Oriented Computing.Cham:Springer,2022:327-333. [10]GRHOS M,ABB L,ELSAYED N,et al.Large language models can accomplish business process management tasks[C]//International Conference on Business Process Management.Cham:Springer,2023:453-465. [11]LASHKEVICH K,MILANI F,AVRAMENKO M,et al.Llm-assisted optimization of waiting time in business processes:A prompting method[C]//International Conference on Business Process Management.Cham:Springer,2024:474-492. [12]FAHLAND D,FOURNIER F,LIMONAD L,et al.How well can large language models explain business processes?[J].ar-Xiv:2401.12846,2024. [13]BONIFATI A,FLETCHER G,VOIGT H,et al.Queryinggraphs[M].San Rafale:Morgan & Claypool Publishers,2018. [14]KOURANI H H,VAN ZELST S J.POWL:partially orderedworkflow language[C]//International Conference on Business Process Management.Cham:Springer,2023:92-108. [15]COURTNEY H,KIRKLAND J,VIGUERIE P.Strategy underuncertainty[J].Harvard Business Review,1997,75(6):67-79. [16]POLYVYANYY A,VANHATALO J,VOLZER H.Simplified computation and generalization of the refined process structure tree[C]//WS-FM 2010.Berlin:Springer,2011:25-41. [17]MARVIN G,HELLEN N,JJINGO D,et al.Prompt engineering in large language models[C]//International Conference on Data Intelligence and Cognitive Informatics.Singapore:Springer,2023:387-402. [18]NI J,ABREGO G H,CONSTANT N,et al.Sentence-t5:Scalable sentence encoders from pre-trained text-to-text models[J].arXiv:2108.08877,2021. [19]KONG A,ZHAO S,CHEN H,et al.Better zero-shot reasoning with role-play prompting[J].arXiv:2308.07702,2023. [20]MARTINO A,IANNELLI M,TRUONG C.Knowledge Injection to Counter Large Language Model Hallucination[C]//European Semantic Web Conference.Cham:Springer,2023:182-185. [21]BROWN T,MANN B,RYDER N,et al.Language models are few-shot learners[J].Advances in Neural Information Proces-sing Systems,2020,33:1877-1901. [22]HUANG L,YU W,MA W,et al.A survey on hallucination in large language models:Principles,taxonomy,challenges,and open questions[J].ACM Transactions on Information Systems,2023,43(2):1-55. [23]HUTH M,RYAN M.Logic in Computer Science:Modelling and reasoning about systems[M].Cambridge:Cambridge University Press,2004. [24]WANG S H,WEN L J,WEI D S,et al.SSDT matrix-based behavioral similarity algorithm for process models[J].Computer Integrated Manufacturing Systems,2013,19(8):1822-1831. [25]ESPARZA J,ROMER S,VOGLER W.An improvement of McMillan’s unfolding algorithm[J].Formal Methods in System Design,2002,20:285-310. [26]SCHAFER B,VAN DER A H,LEOPOLD H,et al.Sketchprocess:End-to-end BPMN sketch recognition based on neural networks[J].IEEE Transactions on Software Engineering,2022,49(4):2621-2641. [27]HU G,WU B,CHEN J.Dynamic adaptation of business process based on context changes:a rule-oriented approach[C]//Ser-vice-Oriented Computing-ICSOC 2013 Workshops.2014:492-504. [28]VASILECAS O,KALIBATIENE D,LAVBIC D.Rule-and context-based dynamic business process modelling and simulation[J].Journal of Systems and Software,2016,122:1-15. |
|
||
首页