TLA Research Journal

Data Analysis for Waste Classification Using Image Processing Techniques of the Office of Academic Resources, Prince of Songkla University

Komgrit Rumdon — Tue, 03 Mar 2026 00:00:00 +0700

This research aims (1) to analyze data for classification of waste images using image processing techniques and (2) to measure the efficiency and accuracy of waste image classification using image processing techniques using the Teachable Machine tool that uses the principles of Machine Learning (ML) to automatically classify waste images. The data used were 250 of 2D vector images of waste, divided into 200 images for learning (Train set) and 50 images for testing (Test set) in a ratio of 80:20. The results of the accuracy measurement found that the overall accuracy was 95.72%. When classifying the accuracy of each type of waste image, it was found that the hazardous waste image had the highest accuracy (99.10%), followed by infectious waste (96.60%), general waste (95.30%), recycled waste (95.10%), and organic waste had the lowest accuracy (92.50%). Data analysis for waste classification using image processing techniques can be developed into a waste sorting system application for use in libraries and various agencies, including for using innovations for energy and environmental conservation activities and supporting the operations of the Green Office and Green Library in using tools to provide knowledge about waste classification to people in the community to be aware of effective waste management.

A Data-Driven Analysis of Library User Segmentation through the CRISP-DM Framework

Pandaree Soonthonwarapas — Tue, 03 Mar 2026 00:00:00 +0700

The purpose of this study was to analyze and segment library users using a data-driven approach by applying the Cross-Industry Standard Process for Data Mining (CRISP-DM) framework. The CRISP-DM framework consists of six phases: business understanding, data understanding, data preparation, modeling, evaluation, and deployment. The dataset used in this study comprises 35,242 transaction records of information resource borrowing from the Khunying Long Athakravisunthorn Learning Resources Center, Prince of Songkla University, Hat Yai Campus, covering the period from January 1 to December 31, 2024. Data analysis was conducted using KNIME analytics software, employing the RFM segmentation technique, which is based on three key indicators: Recency (the most recent library usage), Frequency (the number of transactions), and Monetary (the volume of borrowed resources). The data were analyzed across defined time intervals to classify library users into distinct segments according to their usage behaviors. The results demonstrate that the proposed approach effectively identifies clear and meaningful user segments, enabling the library to gain deeper insights into usage patterns, needs, and potential value of different user groups. In addition, the segmentation model allows for the identification of users who are likely to continue using library services as well as those at risk of reduced or discontinued usage in the future. The findings support strategic decision-making in library management, particularly in designing targeted activities, developing user-specific services, and enhancing user relationship management. Ultimately, this data-driven user segmentation approach contributes to improving service effectiveness, increasing user satisfaction, and sustaining long-term library user engagement.

Artificial Intelligence and the Enhancement of Library Science and Information Science: Potential for Sustainable Development

Yuttana Jaroenruen — Tue, 03 Mar 2026 00:00:00 +0700

In the era of digital revolution, where Artificial Intelligence (AI) technology plays a crucial role across all sectors, libraries and library science cannot avoid this major transformation. This article aims to analyze the role of AI technology in transforming the structure and operational models of modern libraries, as well as its connection to the Sustainable Development Goals (SDGs), particularly SDG 4 on quality education and lifelong learning, through a literature review and analysis of global trends.

The study reveals that modern libraries have undergone a significant transformation in their role from single information service providers to integrated information service hubs. AI technology plays a crucial role in developing services across six main areas: 1) automated conversational systems and virtual assistants that provide 24/7 question-answering and user assistance services; 2) intelligent recommendation and discovery systems that analyze user behavior to suggest resources tailored to individual needs; 3) intelligent information search systems that process natural language and provide highly relevant results; 4) security systems that use image and behavior analysis to prevent resource loss; 5) automated cataloging and classification that enhances collection management efficiency; and 6) resource and inventory management covering demand forecasting, reorder point calculation, and multi-branch resource management.

In education, AI technology has demonstrated potential in supporting SDG4 through the development of adaptive learning systems, academic performance prediction systems, educational resource management systems, and data analytics systems for informed educational decision-making. It also supports lifelong learning by providing age-appropriate content adaptation and personalized learning pathways.

The opportunities and challenges arising from implementing AI technology in library work include the potential for librarians to elevate their roles from information custodians to knowledge managers and learning consultants, allowing for more time to develop creative services, and enhanced collaboration with specialists across various fields. The main challenges include developing technological skills, managing change, addressing privacy and data security concerns, and maintaining balance between increased efficiency and preserving human elements in service delivery.