Intelligent System for Providing Migration Through Dynamic Data Deserialization

Abstract

The purpose of research. Timely provision of data transfer between information systems allows you to quickly exchange resources. However, applications may have different data formats and structures. Therefore, the aim of the research was to develop models, methods and algorithms for a system of asynchronous deserialization of a data string, providing an increase in the efficiency of determining data models by generating strongly typed objects.Methods. The way to deserialize models from data involves line-by-line decomposition of a JSON-file line with the definition of key-value types and their correlation with the data model: character, string, number, boolean value. After that, the web controller conducts asynchronous generation of the class and its objects. To classify string values, serialized string value classifiers are used. For asynchronous generation of objects, a system of “contracts” of models and algorithms for executing and converting these models are used.Results. The deserializer consists of a system of four model analysis controllers and a value generation algorithm. A simple single model deserialization model allows the model to be mapped to relational database table headers to enable model migration between systems. The generated objects are represented by static data types, which ensures that they can be written to any DBMS system using built-in tools. A complex model represents a block of values as a system of different models. Software has been developed for connecting source and target databases, which allows you to migrate data from the created models. Generated values are represented as full-fledged objects and can be used to create a web interface for applications, edit data models, and manage the model system.Conclusion. Experimental studies on deserialization of models from a JSON string containing complex model classes showed an average accuracy of determining the data type of models in 92% of cases, in particular when determining the types of values "character" and "string". The created models are presented in the form of a data table and can be used to ensure the migration of these models.