International Journal of Web Research
Volume:6 Issue: 2, Autumn-Winter 2023

Blockchain is a technology that enables distributed and secure data structures for various business domains. Bitcoin is a notable blockchain application that is a decentralized digital currency with immense popularity and value. Bitcoin involves many concepts and processes that require modelling for better comprehension and development. Modelling is a technique that simplifies and abstracts a system at a certain level of detail and accuracy. Software modelling is applied in Model-Driven Engineering (MDE), which automates the software development process using models and transformations. Domain-specific languages (DSLs) are languages that are customized for a specific domain and offer intuitive syntax for domain experts.  To address the need for specialized tools for Bitcoin blockchain modelling, we propose a novel Unified Modelling Language (UML) profile that is specifically designed for this domain. UML is a standard general-purpose modelling language that can be extended by profiles to support specific domains. A meta-model is a model that defines the syntax and semantics of a modelling language. The proposed meta-model, which includes stereotypes, tagged values, enumerations, and constraints defined by Object Constraint Language (OCL), is defined as a UML profile. The proposed meta-model is implemented in the Sparx Enterprise Architect (Sparx EA) modelling tool, which is a widely used tool for software modelling and design. To validate the practicality and effectiveness of the proposed UML profile, we developed a real-world case study using the proposed meta-model and conducted an evaluation using the Architecture Tradeoff Analysis Method (ATAM). The results showed the proposed UML profile promising.

Given a limited understanding of post-COVID customer behavior on social media, this paper seeks to identify consumer behavioral patterns on Facebook and Instagram in the post-COVID era. The research explores all the inner and outer features that might affect the way consumers think or act in the post-COVID era. Hence, a qualitative approach has been adopted to have a more comprehensive and current understanding. The research begins by identifying pertinent factors shaping consumer behavior through an extensive review of existing literature. To refine and validate these factors, the study employs the fuzzy Delphi method, a research technique that involves seeking consensus among a panel of experts. The findings of the research illuminate 24 distinct factors, organized into five main dimensions: contextual, social, content, technological, and product. Notably, the study underscores the significance of product-related factors, technological advancements, content quality, and social interactions in shaping post-COVID consumer behaviors. The product-related factors hold the most substantial influence over consumer behaviors, followed by technological considerations, content quality, and social interactions. Managers should prioritize quality assurance and strategic product categorization. Emphasizing these aspects in product positioning can significantly influence consumer behavior on social media platforms.

The capital market plays a crucial role within a country's financial structure and is instrumental in funding significant, long-term projects. Investments in the railway transport industry are vital for boosting other economic areas and have a profound impact on macroeconomic dynamics. Nonetheless, the potential for delayed or uncertain returns may deter investors. Accurate predictions of rail company stock prices on exchanges are therefore vital for making informed investment choices and securing sustained investment. This study employs deep learning techniques to forecast the closing prices of MAPNA and Toucaril shares on the Tehran Stock Exchange. It utilizes deep neural networks, specifically One-dimensional Convolutional Neural Networks (1D-CNN), Long Short-Term Memory (LSTM) networks, and a combined CNN-LSTM model, for stock price prediction. The effectiveness of these models is measured using various metrics, including MAE, MSE, RMSE, MAPE, and R2. Findings indicate that deep learning methods can predict stock prices effectively, with the CNN-LSTM model outperforming others in this research. According to the results, The CNN-LSTM model reached the highest R2 of 0.992. Also, based on criteria such as MAE, MSE, RMSE, and MAPE the best results belong to LSTM (Kaggle-modified) with 521.715, 651119.194, 806.920, and 0.028, respectively.

In recent years, determining the most effective time to post on Instagram for increased engagement has become a central concern for digital marketers. Despite its significance, research on this topic remains limited. This study adopts an exploratory research approach to analyze Instagram posts from selected Western countries (Europe and America) and Iranian businesses. The data were collected during the period from February 21, 2022, to March 21, 2022, employing web scraping tools. Classification algorithms, including XGBoost, K-NN, SVM, and Linear Regression, are employed for modeling, with results favoring the XGBoost method for accuracy. The study reveals optimal posting times between 12 and 3 pm for Western Countries businesses and 9 and 12 am for Iranian businesses. Furthermore, it suggests Sunday as the best day for posting in the West, contrasting with Thursday in Iran. In summary, this research underscores the differing ideal posting times in Iran and the West, emphasizing the challenge of constructing a uniform model for all countries.

In order to resolve the issues with Intrusion Detection Systems (IDS), a preprocessing step known as feature selection is utilized. The main objectives of this step are to enhance the accuracy of classification, improve the clustering operation on imbalance dataset and reduce the storage space required. During feature selection, a subset of pertinent and non-duplicative features is chosen from the original set. In this paper, a novel approach for feature selection in intrusion detection is introduced, leveraging an enhanced k-means clustering algorithm. The clustering operation is further improved using the combination of Gravity Search Algorithm (GSA) and Particle Swarm Optimization (PSO) techniques. Additionally, Biogeography Based Optimization (BBO) technique known for its successful performance in addressing classification problems is also employed. To evaluate the proposed approach, it is tested on the UNSW-NB15 intrusion detection dataset. Finally, a comparative analysis is conducted, and the results demonstrate the effectiveness of the proposed approach, in such a way that the value of the detection accuracy parameter in the proposed method was 99.8% and in other methods it was a maximum of 99.2%.

Contemporary machine learning models, like deep neural networks, require substantial labeled datasets for proper training. However, in areas such as natural language processing, a shortage of labeled data can lead to overfitting. To address this challenge, data augmentation, which involves transforming data points to maintain class labels and provide additional valuable information, has become an effective strategy. In this paper, a deep reinforcement learning-based text augmentation method for sentiment analysis was introduced, combining reinforcement learning with deep learning. The technique uses Deep Q-Network (DQN) as the reinforcement learning method to search for an efficient augmentation strategy, employing four text augmentation transformations: random deletion, synonym replacement, random swapping, and random insertion. Additionally, various deep learning networks, including CNN, Bi-LSTM, Transformer, BERT, and XLNet, were evaluated for the training phase. Experimental findings show that the proposed technique can achieve an accuracy of 65.1% with only 20% of the dataset and 69.3% with 40% of the dataset. Furthermore, with just 10% of the dataset, the method yields an F1-score of 62.1%, rising to 69.1% with 40% of the dataset, outperforming previous approaches. Evaluation on the SemEval dataset demonstrates that reinforcement learning can efficiently augment text datasets for improved sentiment analysis results.