Mathematical models of software components for infocommunication service providers

Source abstract

The automation of solving intellectual tasks—specifically, the design of software components for information and communication service providers—using artificial intelligence is an important and timely issue in service lifecycle management platforms for IT development companies based on the End-to-End model. This paper addresses the problem of synthesizing service component code using deep learning models. The requirements for service components are analyzed, and possible service component codes are examined. A mathematical model has been developed to describe the requirements for service software components. The paper analyzes existing mathematical models for representing program code in the Java language that can be used for the mathematical description of service component code. A mathematical model of the program code of service components has been developed, which differs from existing ones in control and adaptation to the features of the grammar of a specific language. This allows for the continued use of unimodal language models. The proposed model is a component of the platform for supporting the life cycle of services in information systems of providers. Within an integrated system of tools for automating service lifecycle processes, the proposed mathematical models will enable the training and implementation of deep neural networks for designing the program code of service components.