CigEngine:基于知识图谱的卷烟数字化设计引擎

CigEngine:a Cigarette Digital Design Engine Based on Knowledge Graph

【目的】卷烟设计流程复杂,所涉及到的数据与算法种类多、结构差异大、关系错综复杂,传统卷烟设计方式存在异构数据及配方管理效率低、算法模型无统一框架支撑等问题,建立卷烟数字化设计引擎可以更好地积累卷烟设计经验,规范设计流程,提升卷烟设计效率。【方法】本文参考借鉴知识图谱相关技术方法,设计并实现卷烟数字化设计引擎CigEngine,CigEngine采用本体对数据、算法、设计流程及配方等进行抽象定义,形成融合“数据-模型-配方”的一体化存储组织模型,并以此为骨架填充具体数据实体与关系,构建卷烟设计知识图谱,卷烟设计流程则以任务的形式由底层分布式计算平台负责调度执行。【结论】Cig Engine充分考虑卷烟设计流程特征,可满足不同品牌不同类型卷烟产品设计需求,目前已驱动构建面向叶组/三丝配方设计、料液/表香设计、烟用辅材设计等应用落地,并取得较好的应用效果。

[Objective]The process of cigarette design is complex,involving various data and algorithms with significant structural disparities and intricate relationships.Traditional design approaches are hindered by inefficiencies in managing heterogeneous data and blends,as well as the absence of a unified framework to support algorithm models.The establishment of a digital cigarette design engine holds the potential to enhance the accumulation of cigarette design experience,standardize the design process,and ultimately improve the efficiency of cigarette design.[Methods]The paper adopts knowledge graph techniques to implement a digital cigarette design engine,CigEngine.CigEngine utilizes ontology to abstractly define data,algorithms,design processes,and blends,forming an integrated“data-model-formula”storage organization model.This ontology model is then used to populate specific data entities and relationships,thereby constructing a knowledge graph for cigarette design.The cigarette design process is scheduled and executed in the form of tasks by a custom-developed distributed computing platform.[Conclusions]CigEngine takes full account of the characteristics of cigarette design processes and can meet the design needs of different brands and types of cigarettes.It has been deployed in multiple tobacco companies and has achieved good application effects.