SPECIFICS OF THE DESIGN AND DEVELOPMENT OF AN INFORMATION SYSTEM FOR DATA PARSING

Abstract

This study proposes and implements an approach to processing images of power cut schedules based on OCR, involving pre-processing of images and post-processing of recognition results (correction of typical errors, time restoration, verification of valid ranges), which improves the suitability of the results for subsequent storage and use in notifications. A modular system architecture has been designed, with components integrated via a centralised PostgreSQL database; a data processing pipeline has also been implemented, ensuring the standardisation of dates and time intervals, data integrity checks, and the prevention of duplicate records during repeated runs and regular monitoring. Particular attention has been paid to ensuring the reliability and integrity of transactions. To this end, the class implements a context manager mechanism, which guarantees the automatic rollback of any changes should errors occur during the execution of operations. Each method of the class performs a preliminary check on the connection status, which prevents disruptions to the operation of asynchronous parsers and the bot in the event of temporary loss of connection to the database server. Using this architecture allows all data manipulation logic to be centralised in a single module, simplifying testing and the further development of the system. The practical significance of this work lies in the creation of a data-parsing information system that reduces the need for manual verification of sources, ensures data is kept up to date during monitoring, and provides users with convenient access to results and alerts. Prospects for the further development of the information system include expanding the list of sources, improving OCR accuracy through adaptive segmentation and additional normalisation rules, enhancing resilience to changes in publication formats, and expanding quality metrics and performance monitoring under real-world operating conditions.