IRIS-Bot: Institutional Research Information System using Intelligent Automation and Deep Learning | IJCT Volume 12 – Issue 6 | IJCT-V12I6P24

The development of the Institutional Research Information System (IRIS-Bot) represents a significant step toward modernizing and streamlining the way academic institutions manage their research outputs. Traditionally, universities rely heavily on manual workflows, decentralized data collection, and inconsistent verification procedures, leading to duplication of effort, inaccurate reporting, and inefficient research tracking. IRIS-Bot directly addresses these challenges through an integrated, automated, and scalable solution. The system successfully combines multiple layers of functionality: automated PDF extraction, metadata enrichment, indexing and citation verification, duplicate detection, domain classification, and semantic search. By leveraging advanced libraries such as PyMuPDF for document parsing, SentenceTransformers for semantic understanding, fuzzy logic for similarity checks, and PostgreSQL with the pgvector extension, IRIS-Bot offers a powerful backend capable of handling large volumes of research data. One of the most impactful contributions of the system is the significant reduction in manual workloads for faculty coordinators and administrative staff. Tasks such as verifying DOI information, checking indexing claims, confirming journal legitimacy, or reviewing citation counts— traditionally time-consuming and error-prone— are now performed automatically and consistently. The system’s enrichment pipeline ensures that data stored in the database is accurate, validated, and complete, allowing institutions to maintain a trustworthy, high-quality research repository. Furthermore, the implementation of a semantic search engine greatly enhances discoverability. This feature is particularly beneficial for scholars seeking to quickly understand the institution’s research strengths and contributions by querying concepts in natural language. The GUI, developed using PySide6, provides a clean, intuitive, and userfriendly way to interact with the system. Overall, IRIS-Bot demonstrates that a thoughtful combination of machine learning, natural language processing, database engineering, and user-centric design can transform institutional research management. The project successfully fulfills its aim of creating a unified, reliable, and intelligent system that simplifies workflows, improves data accuracy, and enhances accessibility. With its modular architecture and extensible pipeline, IRIS-Bot lays a strong foundation for future enhancements and large- scale deployment across departments or institutions.