Abstract:

[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.

Key words: cigarette digital design, ontology model, knowledge graph, workflow