Identification of stakeholder needs and documentation of software requirements are the critical steps to launch a software project. Natural language requirements serve as an agreement among the project stakeholders and they must be transformed into easy-to-understand conceptual models to avoid communication problems. Although conceptual models are mostly created manually with human involvement from the software team, it is seen in recent times that there is a significant increase in studies that automatically generate conceptual models from software requirements. In this study, a Systematic Literature Review (SLR) based on the search of forty-four primary studies (published between 1996 and 2020), which automatically transform software requirements into conceptual models, is reported. These studies are evaluated regarding their approaches, functionalities, dataset used, evaluation methods, generated model types, and languages supported. Finally, several improvable points in the current approaches are highlighted and suggestions are provided as further works.

识别利益相关者的需求并记录软件需求是启动软件项目的关键步骤。自然语言需求是项目利益相关者之间的协议，必须将其转换为易于理解的概念模型，以避免沟通问题。尽管概念模型大部分是在软件团队的人工干预下手动创建的，但是最近发现，可以根据软件需求自动生成概念模型的研究显着增加。在这项研究中，报告了基于对44项主要研究（1996年至2020年之间出版）的搜索的系统文献综述（SLR），该研究自动将软件需求转换为概念模型。对这些研究的方法进行了评估，功能，使用的数据集，评估方法，生成的模型类型和支持的语言。最后，重点介绍了当前方法中的一些可改进之处，并提出了进一步的建议。