Design and Implementation of a Low-Code Intelligent Service Platform

Machine learning algorithms have been used in many applications due to their significant performance, but require a high cost due to the complexity of the implementation process and the high development threshold. In this paper, a novel low-code intelligent service platform (ISP) is proposed, for which the main goal is to reduce the complexity of building models useful for application in different scenarios. Considering the diversity of model development frameworks and programming languages, we have standardized the design of various models into modular components to facilitate their standardized use. To effectively leverage the expertise knowledge and advanced results of model design and development from open source programs on the Internet and the development achievements of co-partners, a component library with various granularity and diverse capabilities can be built across multiple development environments, and some of them will be integrated together by visual arrangement to meet the needs of a wide range of applications. Through component construction, visual arrangement, algorithm reuse, and service decomposition, the platform can provide comprehensive model development and application services for multiple user groups, from beginners to skilled practitioners. Finally, a practical development example of a classification task on police texts is presented. Experimental results show that the ISP, which is capable of developing models with high performance, meets the requirements of model building. Furthermore, the platform, which covers all aspects from data preprocessing to model training and application, provides a reusable, extensible, and visual way to develop modes in different domains.