INTELLIGENT OPEN DATA COLLECTION MODULE IN THE ACADEMIC STAFF PERFORMANCE APPRAISAL SYSTEM

Abstract

The scientific work implements the structure of a module designed for the collection and processing of open data related to scientific works of faculty members at a university, which is incorporated into the performance evaluation system of academic staff. This module serves as part of the current rating system, which is developed using the Laravel framework. The system includes a graphical presentation for users (researchers, heads of departments, reviewers and administrators) and a server-side coded in PHP using JavaScript. A MySQL server that is deployed is utilized for storing the data.
The suggested module automates the gathering of information related to the scientific activity of employees, such as articles, conference papers and other publications, from public sources, including connected external web platforms and publication databases, such as Web of Science, Google Scholar and Scopus. The recognition and categorization of scientific works based on the rating system’s criteria are performed through artificial intelligence, which employs a transformer architecture. This architecture processes and analyzes textual data using a self-attention mechanism, enabling the model to comprehend context, relationships between words, and perform semantic analysis and the process is integrated through the OpenAI API.
For data collection and processing, a Node.js server with Express.js and MongoDB cloud storage via GraphQL API are used. The server integrates with external sources, such as Electronic Library of the National Academy of Educational Sciences of Ukraine and the DSpace. The multi-tiered architecture increases the reliability and scalability of the system. Interaction with the module is provided via JWT Tokens, which guarantees data security and access only to authorized users.