Journal of Soft Computing and Computer Applications

Hate speech detection is crucial as social media diversifies. This research present a lightweight, scalable system using traditional machine learning methods along with a new approach called Spiral-Grey Wolf Optimizer (S-GWO).
S-GWO effectively selects key features that consider both meaning and content from the Term Frequency Inverse Document Frequency (TF-IDF) space, leading to high-quality representation without excessive computing power.
The proposed system was tested on Arabic and another English datasets using six machine learning methods: SVM, RF, LR, KNN, NB, and SGD. It شchieved 92% accuracy and F1 score on the Arabic dataset, while reaching 100% accuracy on the English dataset, significantly reducing
hate speech and toxicity.
Overall, the enhanced algorithms improve accuracy and efficiency, offering an effective alternative to costly deep learning models even with noisy and unbalanced data.

The growing prevalence of cyber threats, including fraud and attacks, has intensified the demand for secure methods of safeguarding confidential information exchanged between users. As telecommunications increasingly rely on multimedia data, video steganography has become
a prominent technique to address these concerns. By embedding sensitive data within video files, this approach enhances protection against unauthorized access and common internet-based attacks, offering
a robust layer of security in an era of escalating digital risks. With the introduction of Deep Learning (DL) steganography methods recently, video steganography can be defined as a rapidly developing subject within information security. This study provides a thorough analysis of the many DL-based video steganography methods that have been covered in the literature. Those methods usually have two modules: a decoder and an encoder. Those techniques have assisted in resolving issues with visual quality as well as capacity, which have been previously evident in older techniques. The study also presented a description of the main databases used in DL-based video steganography, where the University of Central Florida 101 (UCF101) database is the most used in this field. Mean Squared Error (MSE), Peak-Signal Noise-Ratio (PSNR), Average Pixel Di
erence (APD), and Structural Similarity Index (SSIM) have been the top-rated evaluation metrics utilized to assess the ecacy of the video steganography method. All things considered, this paper bridged the gap between two research domains, video steganography and DL, and was a useful tool for academics studying the subject. Previous studies have demonstrated that Reversible Neural Networks (RNNs) and Generative Adversarial Networks (GANs), which rely on Convolutional Neural Networks (CNNs), have given high results.

Video classification is a vital area of research due to the growing volume of video content in various applications. Accurate category across various resolutions poses challenges, which include adapting to scaling, resizing, and compression. Therefore, this paper introduces an innovative Generative Convolutional Network (GCN) set of rules tailored for multi-resolution video classes. The proposed GCN model utilizes Convolutional Neural Networks (CNNs) combined with generative modeling to enhance the extraction of functions across varying video resolutions, which is crucial for maintaining class robustness in the face of common video adjustments, such as scaling, resizing, and compression. In contrast, traditional fashions frequently struggle with such versions, resulting in inconsistent class results. By integrating CNNs with a generative community and utilizing antagonistic training, the GCN renews its function, becoming familiar with and improving its typical class performance. The generative aspect simulates input variations, enabling the model to remain resilient under various video conditions. Experimental results show that the GCN outperforms traditional strategies across key assessment
metrics. On the UCF101 dataset, it achieved 98% accuracy, 94% Recall, 95% precision, and a 96% F-score. Compared to setting up models using SVM, CNN, KNN, and RNN, the GCN can consistently deliver advanced performance, particularly in terms of robustness, accuracy, and resistance to parameter modifications. These findings highlight the potential of GCN to become a new benchmark for video classification, particularly in real-time applications that require dependable evaluation of multi-resolution video content.

Financial and commercial institutions increasingly rely on Extensible Markup Language (XML) files as a standard means of exchanging data. However, this extensive use has created serious security challenges due to the fact that these files contain sensitive information such as bank card
numbers and expiration dates. Relying on traditional full file encryption methods achieves a high degree of security, but it causes problems related to the large file sizes that consume memory and the long encryption and decryption times, which reduces the efficiency of systems when dealing with a large number of daily transactions. Methods based on Type-1 Fuzzy Logic System (T1FLS) have failed to address the ambiguity and uncertainty inherent in data, which can lead to the misclassification of some sensitive elements. Based on this gap, this research proposes an
intelligent model based on Type-2 Fuzzy Logic System (T2FLS) to classify XML file components and determine their security importance more accurately. Partial encryption can then be applied only to the parts classified as highly sensitive. The proposed model was evaluated through a series of experiments on financial files in XML formats. The results showed that it reduced encryption time by 49% compared to full encryption, reduced the size of the encrypted file by approximately 60%, and reduced memory consumption by 32%. The model also demonstrated
strong resistance to common security attacks. This research confirms that combining type-2 fuzzy logic with standard cryptographic techniques achieves a balance between eefficiency and security, making it suitable for application in banking and e-commerce systems.

In agricultural supply chains, the complexity and indeterminacy pose serious challenges to traceability, reliability and confidence today. This challenge is especially acute in the olive oil industry where adulteration, wrong labeling, and uneven chemical quality threaten the actual well-being of producers and consumers. The project aims to design a blockchain-based hybrid architecture with intelligent agents (FNNs) to enhance transparency, reliability and responsiveness in the olive oil supply chain. The Blockchain component enables a completely open, tamper-proof ledger to be built in a very decentralized way and preserved as an archive of every account of its transactions. The intelligent agents contribute to automated decision making, quality control and logistical optimization. The system leverages a real dataset of chemical olive oil with 572 samples with nine fatty acid statistics by being trained, and the system consists of four intelligent agents who work with a specific FNN-based FNN model. The system was validated using agent-based simulations that can reach accuracy levels of 82%, with low error rates, minimum Mean Squared Error (MSE) equal to 0.0157, low execution delay, and no real-world use was possible, the experiments were conducted as controlled simulations, and the feasibility of this model into real supply chain was assessed based on future studies. Results indicated that the computational overhead of the agents is uniform, and the performance is better than that of the ordinary or independent method. This system validates the system real-time and provides reliable information throughout the supply chain, creating an ability to establish real value from a single supplier and in achieving the goal of the SCM in the process of sustainable SCM for the other perishable goods sections.

Hand gesture recognition is a challenging problem in computer vision, particularly in terms of security surveillance applications. This study presents the first efficient system for abduction-related hand gesture real-time detection based on deep learning. The most critical problem is
to detect and recognize hand gestures in real surveillance conditions and to be computationally effective for real-time multi-hand tracking in various lighting situations while allowing reliable surveillance beyond the 1–4 meters limitation. The proposed system consists of three main parts:
The adaptive hand tracking algorithm, which has been used to create the Abductees-Rescue dataset. Introduced pose estimation You Only Look Once version eight, the MediaPipe Hands framework, which variable 3D landmarks extraction at distance. Adaptive Long Short-Term Memory, which was produced by an optimized LSTM model. It has been implemented with the help of the multi-threaded notication approach. All three parts are combined to create novel approaches to the field of abduction rescue research. The Abductees-Rescue dataset includes
9,111 samples, 4,545 native and 4,566 abduction-related 3D hand landmarks. The optimized LSTM model multiplies into a hierarchical signal ow with L2-regularization and an additional dropout approach of achieving 96.12% classifying eciency. Multi-threaded notication has been implemented, allowing for more than two hand-tracking at the same time, all for being 15 FPS sometimes. Pose estimation has been adapted to temporal detection that allows 10-meter-radius action from perpetrators. It utilizes the MediaPipe Hands framework adapted to 3D landmark actions
in the 10 meters distance. Integrated with the Telegram Bot API to enable the multi-threaded engineers to send messages with a picture of abduction delivery time, 1.839 seconds delivery time. This system is a signicant advance in building surveillance-based gesture recognition for
assurance.