MODELS AND SOFTWARE FOR SCIENTIFIC AND TECHNICAL INFORMATION SEARCH

Abstract

This article addresses the problem of efficient search, filtering, and systematization of scientific and technical information by developing an integrated software tool based on modern information retrieval models. The work explores the structure and functionality of leading scientometric databases such as Scopus, Web of Science and discusses their indexing capabilities, metadata handling, and citation-based metrics. Classical information retrieval models, including Boolean, vector space, and probabilistic models, are reviewed along with advanced ranking methods such as BM25. The system also integrates techniques such as topic modeling, clustering, and citation graph analysis to enhance search quality and result relevance. The core contribution of the paper is the design and implementation of a modular web-based information retrieval system that collects metadata and full-text content from major academic publishers Elsevier and IEEE. The system allows for keyword-based searching with filtering options (e.g., date range, journal name, content type), and supports exporting results in BibTeX format for use in reference management. The backend is implemented using PHP and the Laravel framework. The frontend leverages HTML, CSS, and JavaScript for user interaction. A MySQL database ensures structured data storage and fast query execution. The system features three functional modules: data collection (up to five years of publications), periodic data updating via scheduled tasks, and a flexible article search and export interface. Testing confirms the system's ability to streamline the discovery of relevant scientific content and improve the efficiency of research workflows. The proposed tool is particularly useful for researchers, graduate students, and librarians involved in literature reviews, citation analysis, and scientific writing.