{"title":"A systematic review of machine learning for hybrid intelligence in production management","authors":"Carl René Sauer, Peter Burggräf, Fabian Steinberg","doi":"10.1016/j.dajour.2025.100574","DOIUrl":null,"url":null,"abstract":"<div><div>The increasing use of intelligent data processing and its capacity to handle vast data sets enhance efficiency and effectiveness in production management. Consequently, machine learning models have become essential for decision-making in this domain. Previous literature reviews have not considered the perspective of real business requirements from the domain environment, including a knowledge base of theoretical foundations and available methods within the domain. To provide a scientific overview of the current state of the art and to establish a starting point for developing new approaches, this paper presents the results of a systematic literature review. 217 publications were analyzed and synthesized. The publications are classified based on a developed framework that considers the decision type, the production management application, the underlying objective, type, technique, concrete algorithm of the ML model, and decision support for production management issues. A descriptive analysis reveals that there are approaches for all decision types, including unstructured decisions. Surprisingly, some of these approaches are not solely based on simulations to find an optimum. Remarkably, the number of publications related to the type of decision support does not decrease with increasing complexity. Although this paper provides practical guidance to practitioners in selecting applications and ML models to assist their decisions in their production environment, there is a significant need for further research to assist production managers. This can be achieved by developing hybrid models involving interaction between machine and human agents.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"15 ","pages":"Article 100574"},"PeriodicalIF":0.0000,"publicationDate":"2025-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Decision Analytics Journal","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S277266222500030X","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
The increasing use of intelligent data processing and its capacity to handle vast data sets enhance efficiency and effectiveness in production management. Consequently, machine learning models have become essential for decision-making in this domain. Previous literature reviews have not considered the perspective of real business requirements from the domain environment, including a knowledge base of theoretical foundations and available methods within the domain. To provide a scientific overview of the current state of the art and to establish a starting point for developing new approaches, this paper presents the results of a systematic literature review. 217 publications were analyzed and synthesized. The publications are classified based on a developed framework that considers the decision type, the production management application, the underlying objective, type, technique, concrete algorithm of the ML model, and decision support for production management issues. A descriptive analysis reveals that there are approaches for all decision types, including unstructured decisions. Surprisingly, some of these approaches are not solely based on simulations to find an optimum. Remarkably, the number of publications related to the type of decision support does not decrease with increasing complexity. Although this paper provides practical guidance to practitioners in selecting applications and ML models to assist their decisions in their production environment, there is a significant need for further research to assist production managers. This can be achieved by developing hybrid models involving interaction between machine and human agents.