International Journal of Computer Science and Technology

International Journal of Computer Science and Technology | Vol. 3, No. 4, April 2022 | pp. 25–32

Research Article

Title: Improving Aspect-Based Sentiment Analysis in E-Commerce Reviews Using Multi-Head Attention Graph Convolutional Networks

Names of Authors: Siti Aminah¹, Budi Santoso², and Agus Wijaya³

Authors’ Affiliations:
¹Department of Computer Science and Information Systems, Universitas Indonesia, Jakarta, Indonesia
²School of Electrical Engineering and Informatics, Bandung Institute of Technology, Bandung, Indonesia
³Department of Informatics, Sepuluh Nopember Institute of Technology, Surabaya, Indonesia

Abstract: Aspect-Based Sentiment Analysis (ABSA) is a vital text-mining task that enables e-commerce companies to extract detailed consumer feedback by identifying specific product aspects and their associated emotional tones. However, context sentences often present complex grammatical structures and multi-word feature expressions that distance an aspect from its descriptor words, confusing standard recurrent neural network models. This study proposes a novel ABSA framework that utilizes a multi-head self-attention mechanism integrated with a Graph Convolutional Network (GCN). The system uses a dependency parser to convert an input text review into a syntactic dependency graph, mapping structural links between individual tokens. The GCN layer operates directly on this graph structure, extracting syntax-aware word features, while the multi-head self-attention module captures long-range semantic dependencies across the sentence layout. Experiments conducted on standard consumer electronic evaluation datasets demonstrate that the proposed attention-driven GCN model achieves an overall classification accuracy of 86.4% and an F1-score of 79.8%. These metrics represent a significant performance gain over baseline LSTM and standard BERT sequence classification systems, demonstrating high reliability in extracting consumer sentiments from complex, informal text structures.

Keywords: Aspect-based sentiment analysis, Natural language processing, Graph convolutional networks, Attention mechanisms, Text mining, E-commerce feedback

Manuscript Timeline: Received: January 14, 2022; Revised: February 22, 2022; Accepted: March 18, 2022; Published: April 1, 2022

Citation: Aminah, S., Santoso, B., & Wijaya, A. (2022). Improving aspect-based sentiment analysis in e-commerce reviews using multi-head attention graph convolutional networks. International Journal of Computer Science and Technology, 3(4), 25–32. DOI: 10.46882/2022/IJCST/000028

International Journal of Computer Science and Technology | Vol. 3, No. 3, March 2022 | pp. 17–24

Research Article

Title: An Energy-Aware Virtual Machine Consolidation Framework for Cloud Datacenters Using Deep Reinforcement Learning

Names of Authors: Linus Bergqvist¹, Elsa Lindstrom², and Oscar Nilsson³

Authors’ Affiliations:
¹Department of Computer Science and Engineering, Chalmers University of Technology, Gothenburg, Sweden
²School of Electrical Engineering and Computer Science, KTH Royal Institute of Technology, Stockholm, Sweden
³Department of Information Technology, Uppsala University, Uppsala, Sweden

Abstract: Cloud computing centers generate substantial carbon footprints and incur high operational expenditures due to the continuous underutilization of physical server hardware. Dynamic virtual machine (VM) consolidation reduces energy consumption by migrating active VMs out of underutilized host nodes, allowing idle physical hardware to transition into low-power sleep modes. However, excessive VM migration activities can cause severe host resource contention and induce service level agreement (SLA) performance degradation. This paper presents an energy-aware VM consolidation framework that utilizes a deep reinforcement learning architecture based on the Asynchronous Advantage Actor-Critic (A3C) algorithm. The consolidation task is formulated as a Markov decision process where state signals capture host CPU loads, RAM utilization parameters, and network input-output traffic. The A3C neural networks learn an optimal consolidation policy that dynamically triggers VM migrations, balancing host energy savings against potential SLA violations. Simulations using standard cloud workload traces show that the proposed reinforcement learning framework cuts overall datacenter power consumption by 24.3% compared to static threshold-based migration approaches. Crucially, total VM migration counts drop by 31.5%, maintaining system performance stability under highly fluctuating traffic conditions.

Keywords: Cloud computing, Virtual machine consolidation, Deep reinforcement learning, Energy efficiency, Green computing, Resource orchestration

Manuscript Timeline: Received: December 08, 2021; Revised: January 19, 2022; Accepted: February 14, 2022; Published: March 1, 2022

Citation: Bergqvist, L., Lindstrom, E., & Nilsson, O. (2022). An energy-aware virtual machine consolidation framework for cloud datacenters using deep reinforcement learning. International Journal of Computer Science and Technology, 3(3), 17–24. DOI: 10.46882/2022/IJCST/000027

International Journal of Computer Science and Technology | Vol. 3, No. 2, February 2022 | pp. 9–16

Research Article

Title: A Verifiable Multi-Signature Scheme for Secure and Auditable Supply Chain Tracking Systems

Names of Authors: Marta Wisniewska¹, Jan Kowalski², and Piotr Nowak³

Authors’ Affiliations:
¹Faculty of Electronics and Information Technology, Warsaw University of Technology, Warsaw, Poland
²Institute of Computer Science, Jagiellonian University, Kraków, Poland
³Department of Cyber Security, AGH University of Science and Technology, Kraków, Poland

Abstract: Global supply chains face escalating threats from product counterfeiting, regulatory non-compliance, and unauthorized cargo tampering during transit across international logistics hubs. Traditional database tracking mechanisms lack cross-organizational transparency, allowing malicious actors to alter shipping manifests or forge certification records. This paper introduces a cryptographically secure, auditable supply chain tracking protocol that utilizes a verifiable multi-signature scheme based on bilinear pairings. The proposed protocol compresses verification signatures from multiple transit entities—such as manufacturers, logistics firms, customs agents, and retailers—into a single, compact cryptographic proof string. This multi-signature proof is recorded directly onto a distributed ledger network, ensuring immutable tracking and non-repudiation without overloading ledger storage space. The system implements a threshold validation mechanism, requiring a minimum number of valid participant signatures before a product shipment updates its custody status on-chain. Cryptographic benchmarks show that signature generation requires 2.4 ms per participant, and the combined verification loop processes in under 5.1 ms. The security analysis demonstrates that the protocol achieves existential unforgeability under chosen-ciphertext conditions, providing a robust security framework for global tracking systems.

Keywords: Cryptographic protocols, Multi-signatures, Supply chain security, Distributed ledgers, Bilinear pairings, Data auditability

Manuscript Timeline: Received: November 12, 2021; Revised: December 18, 2021; Accepted: January 15, 2022定位; Published: February 1, 2022

Citation: Wisniewska, M., Kowalski, J., & Nowak, P. (2022). A verifiable multi-signature scheme for secure and auditable supply chain tracking systems. International Journal of Computer Science and Technology, 3(2), 9–16. DOI: 10.46882/2022/IJCST/000026

International Journal of Computer Science and Technology | Vol. 3, No. 1, January 2022 | pp. 1–8

Research Article

Title: An Edge-Assisted Computer Vision System for Automated Safety Compliance Monitoring in Industrial Facilities

Names of Authors: Diego Silva¹, Mateo Fernandez², and Isabella Gomez³

Authors’ Affiliations:
¹Department of Automation and Robotics, Universidad de Chile, Santiago, Chile
²Faculty of Engineering, Universidad de los Andes, Bogotá, Colombia
³School of Electrical Engineering, Universidad Nacional Autónoma de México, Mexico City, Mexico

Abstract: Maintaining strict occupational safety compliance, including the proper use of personal protective equipment (PPE) like hard hats and safety vests, is essential for reducing workplace injury rates in heavy industrial plants. Manual compliance auditing by floor supervisors is labor-intensive, intermittent, and prone to oversight, making continuous, automated monitoring solutions highly desirable. This paper presents an edge-assisted computer vision framework designed for real-time PPE detection and safety perimeter monitoring. The core architecture uses an optimized version of the YOLOv5 object detection network, enhanced with depthwise separable convolutions to minimize computational demands. The model processes high-resolution industrial video feeds on localized edge gateways, identifying safety gear compliance and warning when personnel enter hazardous operational zones. To improve detection under dust and variable lighting conditions, a spatial pyramid pooling module is integrated into the model pipeline. Field testing on an industrial processing floor demonstrates a mean average precision (mAP) of 94.6% for hard hat and vest classification. The localized edge processors achieve an inference speed of 32 ms per frame, allowing immediate audio-visual alarms during safety breaches while saving wide-area network bandwidth by avoiding cloud video streaming.

Keywords: Computer vision, Object detection, Industrial safety, Edge computing, Deep learning, Real-time monitoring

Manuscript Timeline: Received: October 14, 2021; Revised: November 20, 2021; Accepted: December 12, 2021; Published: January 4, 2022

Citation: Silva, D., Fernandez, M., & Gomez, I. (2022). An edge-assisted computer vision system for automated safety compliance monitoring in industrial facilities. International Journal of Computer Science and Technology, 3(1), 1–8. DOI: 10.46882/2022/IJCST/000025

International Journal of Computer Science and Technology | Vol. 2, No. 12, December 2021 | pp. 89–96

Research Article

Title: A Multi-Criteria Task Scheduling Framework for Heterogeneous Cloud Systems Using Simulated Annealing and Ant Colony Optimization

Names of Authors: Kofi Mensah¹, Kwame Boateng², and Ama Asare³

Authors’ Affiliations:
¹Department of Computer Science, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana
²Department of Computer Engineering, University of Ghana, Accra, Ghana
³School of Technology, GIMPA, Accra, Ghana

Abstract: Efficient task scheduling remains a critical operational priority for infrastructure-as-a-service cloud datacenters seeking to optimize system performance while reducing resource overheads. The assignment of diverse, inter-dependent computational tasks onto non-uniform physical servers represents a complex, multi-objective NP-hard problem. Standard scheduling heuristics usually prioritize single metrics like overall completion time, which frequently leads to uneven server load distribution, high energy waste, and service level agreement violations. This study proposes an optimized hybrid metaheuristic framework, SA-ACO, combining Ant Colony Optimization (ACO) with Simulated Annealing (SA) to resolve complex cloud scheduling demands. The algorithm utilizes an initial ACO loop to establish global routing routes across available nodes, while an integrated SA mutation routine alters local node assignments to avoid getting trapped in local optima. The objective model evaluates multiple performance trade-offs, factoring in task completion time, resource utilization balance, and server energy consumption. Extensive benchmark simulations with 1000 tasks running across 100 virtual machines reveal that the SA-ACO framework lowers overall completion time by 21.4% and decreases system energy consumption by 18.7% compared to traditional genetic algorithms and standard round-robin scheduling setups.

Keywords: Cloud computing, Task scheduling, Ant colony optimization, Simulated annealing, Resource allocation, Energy optimization

Manuscript Timeline: Received: September 15, 2021; Revised: October 24, 2021; Accepted: November 19, 2021; Published: December 1, 2021

Citation: Mensah, K., Boateng, K., & Asare, A. (2021). A multi-criteria task scheduling framework for heterogeneous cloud systems using simulated annealing and ant colony optimization. International Journal of Computer Science and Technology, 2(12), 89–96. DOI: 10.46882/2021/IJCST/000024

International Journal of Computer Science and Technology | Vol. 2, No. 11, November 2021 | pp. 81–88

Research Article

Title: An Automated Malicious Domain Detection Architecture Using Recurrent Neural Networks and Passive DNS Analysis

Names of Authors: Youssef Mansour¹, Hassan Farhat², and Fatima Al-Sayed³

Authors’ Affiliations:
¹Department of Computer Science, American University of Beirut, Beirut, Lebanon
²Faculty of Technology, Lebanese University, Sidon, Lebanon
³Department of Computer Engineering, Beirut Arab University, Beirut, Lebanon

Abstract: Cybercriminals regularly utilize algorithmic techniques like Domain Generation Algorithms (DGAs) to generate vast lists of temporary internet domains for command-and-control communication. Because these domains change rapidly, conventional reputation blacklists struggle to provide real-time protection, leaving enterprise networks vulnerable to ransomware and data extraction campaigns. This paper introduces an automated, high-precision malicious domain detection system that combines structural character analysis with passive DNS traffic modeling. The core architecture implements a long short-term memory (LSTM) recurrent neural network that checks string sequences to flag algorithmically generated domains before they receive active network requests. Simultaneously, a statistical profiling layer monitors passive DNS telemetry, assessing attributes like query frequency variations, geographic IP distributions, and resource record changes. Training and validation leverage a large dataset of 500,000 active domains, spanning both legitimate entries and real-world botnet traffic. Experimental metrics demonstrate an overall detection accuracy of 98.9% alongside a low false positive rate of 0.014%. The integrated detection engine processes incoming queries in less than 3.5 ms, making it highly effective for deployment within active corporate DNS gateways to block zero-day command-and-control communication channels.

Keywords: Network security, Domain generation algorithms, Long short-term memory, Passive DNS, Intrusion prevention, Cyber forensics

Manuscript Timeline: Received: August 18, 2021; Revised: September 22, 2021; Accepted: October 14, 2021; Published: November 1, 2021

Citation: Mansour, Y., Farhat, H., & Al-Sayed, F. (2021). An automated malicious domain detection architecture using recurrent neural networks and passive DNS analysis. International Journal of Computer Science and Technology, 2(11), 81–88. DOI: 10.46882/2021/IJCST/000023