Header Journal

Albaha University Journal of Basic and Applied Sciences

p-ISSN: 1658-7529

e-ISSN: 1658-7537

Towards emotion-photos classification in social networks according to their context and visual features

Towards emotion-photos classification in social networks according to their context and visual features
  • Sonia LAJMI
  • Reseived: Received in Revised Form: 12 January, 2025 Accepted: 18 January, 2025
  • 17 August, 2024

Online Social Networks have become increasingly popular as platforms where individuals can express their emotions, thoughts, and opinions through the use of photos and comments. Therefore, the significant rise in the number of social photos being shared online has become an opportunity for the exploration of comprehensive photo analysis. This study focuses on the classification of social photos according to their emotional content. Current studies in this area have primarily focused on determining the sentiment's polarity, specifically whether it is positive or negative. This study presents a new approach for classifying emotions in social photos. Various criteria are used to classify tailored photo emotions, including visual aspects and social context. The combination of these factors is consistently used to create a system for classifying emotional photos.

download

The Effect of Campus Design on Urban Development in Saudi Arabia: A Case Study of Al-Baha University

The Effect of Campus Design on Urban Development in Saudi Arabia: A Case Study of Al-Baha University
  • Naief A. Aldossary, Khalid A. Alkhuzai
  • Reseived: 30 October, 2024 Received in Revised Form: 28 January, 2025 Accepted: 10 February, 2025

The paper investigates and analyses the urban design of the Al-Baha University campus from an economic and environmental perspective. With coronavirus pandemic, the education protocol, precautionary measures and social distance have been applied in order to achieve safety environment within the university boundary. The study analyses the university campus design in light of sustainable requirements, social distance protocol, employment of natural resources and application of urban sustainable design criteria. Regional urban planning and development needs to include attraction factors to boost the economy and trigger reverse migration. Establishing a university in the region helps significantly in terms of local urban growth but environmental aspects including energy conservation and management should also be considered for environmental reasons. The approach used in this study involved: (a) an environmental site analysis including an analysis of natural resources such as the wind, solar energy and annual rainfall to determine how such resources can be employed in the operation of the university; (b) a site visit survey analysis to investigate the sustainable use of the current facilities and open spaces of the university campus and the application of on-site renewable energy; and (c) a SWOT analysis to determine the role of the university in the local economy system. The urban sustainable design and possibility of using landscaping, shading devices, and solar radiation systems was investigated through building plans and a site visit to the campus. In addition, the form and fabric of the academic buildings was investigated to identify the application of shading devices, the windows design and use of insulation. The study highlights the fact that many buildings are still under construction while other buildings are still currently on plan. The paper concludes by giving some recommendations from an economic and environmental perspective.

download

Records of Dysphania pumilio and Sclerolaena birchii (Amaranthaceae) as new alien species in the flora of Al-Baha region, Saudi Arabia

Records of Dysphania pumilio and Sclerolaena birchii (Amaranthaceae) as new alien species in the flora of Al-Baha region, Saudi Arabia.
  • Nageeb A. Al-Sagheer
  • Reseived: 26 January, 2025 Received in Revised Form: 3 March, 2025 Accepted: 16 April, 2025

The findings were part of a regular inventory conducted in the mountain forest areas of both Al-Baha and Asir regions during the third and fourth quarters of 2024. The species were recorded within randomly established quadrats laid out to survey plant diversity and species richness. Intensive and regular fieldwork, using 20 x 20 m random quadrats, was carried out along the high-altitude strips stretching from Al-Mandak to Baljurashi. This effort led to the identification of Dysphania pumilio (R. Br.) Mosyakin and Clemants and Sclerolaena birchii (F.Muell.) Domin of the Amaranthaceae family as invasive plant species within the flora of Saudi Arabia. Among the associated species, the alien plants Sclerolaena birchii and Dysphania pumilio exhibited an increase in both density and frequency. Although Sclerolaena birchii was found in only one location, it ranked ninth in density, with a value of 0.025 per hectare and a frequency of 25%. In contrast, Dysphania pumilio ranked fifth, showing a higher density of 0.125 per hectare and a frequency of 100%. Alien plant invasions are threatening the species structure and composition of Al-Baha forests due to interactions with various driving factors. To address the issue, strict regulations must be implemented to control the trading and importation of plant parts, including seeds, seedlings, and fruits. Emphasis should be placed on prevention, early detection, and rapid response. Equally important is raising public awareness, encouraging community participation, and advancing scientific research to develop sustainable strategies for managing invasive species. By prioritizing these efforts, authorities can protect native ecosystems and mitigate the adverse impacts of biological invasions

download

Sentiment Analysis with Text Augmentation of YouTube Comments for Arabic Language Learners

Sentiment Analysis with Text Augmentation of YouTube Comments for Arabic Language Learners
  • Maha Alamri
  • Reseived: 20 February, 2025 Received in Revised Form: 1 May, 2025 Accepted: 8 May, 2025

This paper investigates sentiments expressed in YouTube comments made by Arabic language learners. Given its popularity as a video-sharing platform, it allows for a diverse array of educational materials, which can be most useful in language acquisition. Since Arabic is rapidly becoming one of the most spoken languages in the world, it poses unique challenges to learners owing to its grammar, script, and phonetics. In relation to learner engagement and sentiment, this paper presents a manually annotated dataset comprising 10,209 comments from Arabic language learners covering a various range of topics from the Arabic alphabet, greetings, keywords and expressions, script recognition, to reading lessons. In addition, data augmentation techniques have been employed to further enhance the dataset, thus introducing a wider variety of training samples. The analysis involved sentiment classification based on three classifiers: Support Vector Machines (SVM), Multinomial Naive Bayes (MNB), and Random Forest (RF). Experimental results showed that RF achieved the best classification among the algorithms tested.

download

History Analysis of Competing Graduation and Retention Events for Engineering Programs

History Analysis of Competing Graduation and Retention Events for Engineering Programs
  • Ahmed M. Bagabir, Fares S. Alareqi
  • Reseived: 16 May 2024 Received in Revised Form: 31 May 2024 Accepted: 22 April, 2025

Extensive research is required to determine the factors that influence graduation rates and to develop compatible educational policies. This study aims to predict factors related to students and institutions that are associated with graduation rates and the risk of dropping out. The present paper employs the cause-specific hazard (CSH) model of event history analysis (EHA) to analyze competing-risk data. The model focuses on academic terms as the time that a student spends until the completion or termination of their undergraduate program. The analysis is based on a database of 13 cohorts containing 4571 students enrolled in engineering programs at a university in Saudi Arabia. The findings of the model are quite revealing, emphasizing that first-year achievements and factors related to student enrollment are the most significant predictors that can drastically increase the graduation rate and simultaneously reduce the risk of dropping out. The study highlights that students are still vulnerable to dropping out despite the absence of tuition fees and the provision of sustainable financial support to national students. This is because poor academic performance resulting from a student’s cognitive inability to cope with engineering studies is one of the significant reasons for dropouts, and this can severely strain the university’s capabilities for an extended period. Therefore, it is crucial to address these challenges and provide the necessary support to students, especially during their first year, to ensure they achieve academic success and graduate on time. The implications of this research can be valuable to higher education institutions in adjusting admission and study policies and taking initiatives to increase persistence to support student success.

download

Deep Q-Learning Based Method for Denial of Service (DoS) Attack Detection

Deep Q-Learning Based Method for Denial of Service (DoS) Attack Detection
  • Asmaa A.Alghamdi Nizar H.Alsharif
  • Reseived: 17 March, 2025 Received in Revised Form: 10 May, 2025 Accepted: 18 May, 2025

A Denial-of-Service (DoS) attack is a type of cyberattack in which an attacker attempts to overload a website or network with excessive traffic, causing it to become unavailable to legitimate users. The goal of a DoS attack is to disrupt the normal functioning of a website or network, making it inaccessible to legitimate users. To defend against DOS attacks, we propose an approach that uses deep-Q learning to detect more powerful and obfuscating attacks. The proposed approach defending against low-rate Denial-of-Service (DoS) attacks uses deep reinforcement Q-learning as a powerful decision-making tool to detect more powerful and obfuscated attacks. This approach aims to overcome the limitations of signature-based systems by implementing an adaptive learning mechanism that can detect attacks before they affect the network. It improves the system's detection accuracy, making it more effective at defending against low-rate DoS attacks. The experimental results of the detection phase of the proposed approach showed that the method can effectively distinguish abnormal network attack flows with higher detection accuracy than other algorithms. The proposed approach achieves promising performances, i.e.: accuracy of 99.3% with precession, recall and F1-score value is 98.5%, 100%, and 99%, respectively. In addition, the method showed a certain rate of new unknown attack detection, indicating that it is effective and suitable for the actual network environment. These results indicate that the proposed method can improve the accuracy and speed of detecting DoS attacks, which is crucial in defending against them.

download

A Proposed Faults Classification Using Feature Extraction and Long Short-Term Memory Network

A Proposed Faults Classification Using Feature Extraction and Long Short-Term Memory Network
  • Mishari Metab Almalki
  • Reseived: 18 May, 2025 Received in Revised Form: 25 May, 2025 Accepted: 26 May, 2025

Incipient faults (IFs) and high impedance faults (HIFs) plague medium voltage distribution systems. The IFs are self-cleared and occur in underground cables. They may evolve into permanent faults if they were left unrepaired. They are not detectable by conventional relays due to self-clearing phenomenon. Similarly, the HIFs are not detected by traditional relays as well. This is due to the low fault current generated by an energized broken conductor that touching with surface of the ground. This current is similar to large industrial loads. This paper proposes classification process via a feature extraction using time-frequency analysis and Long Short-Term Memory networks. At first, these faults, as well as load switching, capacitor energization and permeant faults (Low Impedance Faults -LIFs-) were created on the IEEE-34 and 13 bus systems. Then, these events were detected. Then, the registered phase current at the substation level for each event is utilized by the classification method to calculates two important characteristics for the phase current signals using time-frequency analysis (Instantaneous Frequency and Spectral Entropy). Both features are used to construct a Feature Matrix for each event. A Long Short-Term Memory (LSTM) Network was trained with feature matrices. The findings showed that the proposed method had an F-1 score of 99.03% classification accuracy for proposed method.

download