Automating the development of semantic models for application programming interfaces

Abstract

This article analyzes the problem of automating the creation of semantic models for application programming interfaces, which provides the transformation of user queries in natural language into accurate API method calls. Existing approaches to developing natural language user interfaces are reviewed, including the use of natural language processing techniques, pre-trained language models, and specialized thesauruses. The paper proposes an algorithm for the automated generation of semantic models for application programming interfaces. A test environment was implemented to evaluate the proposed approach, providing dynamic mapping of user queries to API methods. Based on the results, it can be concluded that the proposed method improves the accuracy and completeness of query transformations, minimizes manual intervention, and ensures the model’s adaptability to changes in API structures.