A Model for Developing My GPTs Using Human-in-the-Loop Approach to Enhance Journal Indexing Efficiency

Abstract

             This study aimed to design a development model for My GPTs based on the Human-in-the-Loop (HITL)concept and develop efficient My GPTs for journal indexing cataloging. The research employed a Research & Development (R&D) approach, applying the PAOR model with the MAPE-K model. The research was conducted in two phases: Phase 1 involved problem analysis and development of a prototype My GPTs as a bibliographic cataloging assistant for journal indexing, and Phase 2 involved testing My GPTs for journal indexing cataloging, improvement based on user feedback, and evaluation. The study population consisted of 8 branch library staff responsible for journal indexing and one librarian from the Information Resources Management department. The research instruments included a bibliographic accuracy verification form and a satisfaction interview form for My GPTs usage. The research results on performance evaluation of My GPTs developed based on the model with HITL concept showed that the accuracy of cataloging entries after using My GPTs increased by 25.44% compared to before usage (from 73.78% to 92.55%), and the overall satisfaction evaluation for My GPTs usage was 76.11%, which is considered excellent. This My GPTs development model can serve as a practical guideline for developing Generative AI for library information resource cataloging.