Development of the Web Application “KMITL SDG Matching for All” for Bilingual Text Analysis to Link with the Sustainable Development Goals (SDGs) using A Hierarchical Reasoning Structure

Abstract

This research, conducted under the Routine to Research (R2R) framework, identifies that currently, there is no system that can accurately analyze Thai-language text to identify its relevance to the Sustainable Development Goals (SDGs). This study aims to develop a web application called “KMITL SDG Matching for All” to analyze both Thai and English texts in order to determine which SDGs the content relates to. The system uses a hierarchical reasoning tree structure based on the keyword set from the Elsevier SDG mapping queries (2023). It can process information from direct keywords to target groups, types of activities, and policy context. The system is designed to be accurate and transparent without relying on external Artificial Intelligence (AI), and therefore has no cost and does not require commercial API connections. The performance evaluation was conducted in two ways: (1) by comparing the accuracy with the keyword-based SDG matching system from Chiang Mai University, and (2) by comparing the analysis results with ChatGPT to assess the similarity to contextual AI. The results show that the system achieved an accuracy of 87.5%, which is higher than the general keyword system, which achieved only 46.0%, and had a consistency rate with ChatGPT at 83.5%. This system has the potential to be applied in higher education institutions for purposes such as public communication and supporting data preparation for international university sustainability rankings.