JOURNAL OF INFORMATION SCIENCE AND TECHNOLOGY

Standing Posture Monitoring System for the Elderly

Patcharaporn Saenkaew — 2024-12-28

The elderly population is increasing rapidly in Thailand, with the majority of seniors living alone at home. Statistics reveal that, on average, four elderly individuals die each day due to falls. This research develops a standing up notification system for the elderly using Raspberry Pi to transmit motion image when they rise from bed. The image is processed using keypoint extraction, and the extracted points are used to verify the correctness of the standing posture by calculating suitable angles from various body parts. The comparison of the detection accuracy of standing postures between two libraries, MediaPipe and OpenPose, using the Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) shows that MediaPipe has lower errors at 25.19 and 20.35, respectively. The accuracy of standing posture detection is further validated by calculating angles in various degrees, revealing an overall highest average accuracy of 74.55%.

Development of a Web Application for Student Navigation and Home Visits

Phuwadon Yubonmat — 2024-12-28

This research presents the development of a web application for student navigation and home visits and evaluates the user satisfaction with the application. The users are divided into three groups: responsible teachers (administrators), homeroom teachers, and students. The system is developed using HTML, CSS, JavaScript, and Leaflet for displaying maps and student home coordinates. The web application manages student and homeroom teacher information, records student home coordinates, searches for routes to student homes, logs home visit information, manages scholarships, and generates home visit reports. The user satisfaction evaluation indicates a high level of satisfaction, with an average score (x̅) of 4.48 and a standard deviation (S.D.) of 0.58.

Thai Language Sentiment Analysis with a Hybrid Method on WangchanBERTa-CNN-BiLSTM

Kasidhit Suraratchai — 2024-12-28

Understanding emotions conveyed in text, especially in non-global languages such as Thai, sentiment analysis is particularly important in Thailand. However, this endeavor faces challenges due to variations in text length, which significantly impact sentiment analysis outcomes. Previous research has employed neural network and machine learning models in the process, yet each model specializes in different aspects, making comprehensive sentiment analysis coverage unattainable. Recent research has delved into hybrid models like CNN-BiLSTM and BiLSTM-CNN. Although they demonstrate efficacy, their performance still varies across different datasets. For instance, CNN-BiLSTM excels with short sentences by considering surrounding word context, while BiLSTM-CNN is more effective with long sentences due to its bidirectional learning capability. While showing promise, these models perform effectively, but varied text lengths in datasets often lead to sentiment misinterpretation. To address these challenges, inspired by recent advances, we propose an innovative solution: the Parallel Hybrid model. This approach integrates WangchanBERTa into both CNN-BiLSTM and BiLSTM-CNN architectures, harnessing ensemble techniques to improve overall performance and adaptability. Our experiments, conducted on datasets like Wisesight, a highly imbalanced dataset with mostly longer texts, and Thai Children's Tales, a less imbalanced dataset with mostly shorter texts, confirm the effectiveness of the Parallel Hybrid model, which outperforms other model configurations with Macro F1 scores of 0.6270 and 0.7859, respectively. This research marks a significant advancement in sentiment analysis for the Thai language.

An Application of Machine Learning Techniques for Loan Default Payment Prediction

Wilawan Inchamnam — 2024-12-28

In the banking business, predicting customer default payments has become a crucial operation to prevent and mitigate risks caused by non-performing loans. Presently, machine learning techniques are used alongside traditional methods for this task. This paper explores several ways to apply machine learning techniques in predicting default payments. The prediction development framework includes data encoding, data sampling, and model development. At each step, various techniques are tested and compared to find optimal solutions for business requirements. Our findings conclude that ensemble models are a good choice over a single model to increase the precision of the default payment class. The Over-sampling method is a suitable choice to increase recall of the default payment class, whereas the Under-sampling method is not recommended. Furthermore, if the size of the input vector is a concern, the Weight of Evidence encoding method can be used instead of One-hot encoding without a loss in performance.

Agriculture Virtual Banking Platform

Nanta Janpitak — 2024-12-28

In Thailand, there has been a growing interest in agriculture among the population, both as a primary occupation and as a hobby. Many individuals engage in small-scale farming, such as growing vegetables for personal consumption or for small-scale trade. A significant obstacle they often encounter is the lack of sufficient equipment or resources. These resources are not always feasible to purchase due to infrequent use or high costs, such as tractors, grass cutters, or battery-powered sprayers. To address this issue, the researchers propose the concept of an Agriculture Virtual Banking Platform inspired by Thailand 4.0 principles. This system facilitates the exchange and lending of agricultural equipment, promoting convenience and strengthening community resilience through the shared use of resources. The Agriculture Virtual Banking Platform is designed to allow community members to exchange or rent agricultural resources. The system functions similarly to a financial bank, where members with abundant resources can establish their own banks to offer rental services. Conversely, members with fewer resources can deposit their (virtual) resources in a nearby bank and earn interest or rental fees when others rent their equipment. This system not only reduces the cost of purchasing agricultural equipment but also extends the usability and durability of these resources. For instance, tools that are rarely used by their owners can be rented out, thereby preventing rust and maximizing their utility. The Agriculture Virtual Banking Platform will promote the development of quality of life and community, fostering tangible self-reliance among farmers and community members.

An intelligent watering system for sweet corn cultivation with LoRa technology

Ponglert Sangkaphet — 2024-12-28

The cultivation of sweet corn is a field that utilizes a vast area, thereby the care and watering of sweet corn plants are inconvenient and time-consuming. Currently, Internet of Things technology plays a significant role in sending and receiving data, contributing to the development of a smart farm system. However, the Internet of Things still has a limitation on the distance over which data cannot be transmitted. Therefore, this research aims to develop a prototype smart watering system for cultivating sweet corn using LoRa technology, which relies on both receiving and sending data from water valve nodes and sensor nodes. These components communicate data in a peer-to-peer format. The results of system development and testing lead to the conclusion that the system can effectively operate the water valves in both on/off modes and successfully transmit values from remote sensors. The research sample consisted of 5 program development experts and 15 general users. The tools utilized in the research included a system performance evaluation form, and the statistics employed for data analysis were the mean and standard deviation. The results of the overall efficiency evaluation were at a high level, with a mean score of 4.43.

Virtual Reality Design Guidelines to Enhance Movement in Elderly

Angkana Suwanjatuporn — 2024-12-28

The objectives of this research are 1) to study the behavior of receiving media from Thailand’s news agencies on Facebook medium to be able to use in the news agency's public relations planning. 2) To study a relationship between preferences for political parties and news agencies on Facebook. The results showed that most of the respondents were female, aged between 26-43 years old, educated at a bachelor's degree. Have a career in government service/state enterprise and average monthly income between 15,001 - 30,000 baht, having a single status and having 3-4 family members. The news and the main occupation had a statistically significant relationship. In the aspect of the hypothesis, the relationship between news agencies on Facebook. With political party preference was found to have a statistically significant relationship.

A Comparative Study of 3D Navigation Techniques Using Trackpads in Desktop VR

Banyapon Poolsawas — 2024-12-28

The limited input capabilities of trackpads compared to traditional mice hinder smooth 3D navigation, impacting productivity and user experience in design and gaming software. This research investigates the effectiveness of current trackpad-based 3D navigation methods. By comparing task performance user feedback, and identifying the strengths and weaknesses of various techniques, the study aims to guide the development of more intuitive trackpad navigation solutions.