In recent years, the importance of computer security has increased due to the rapid advancement of digital technology, widespread Internet use, and increased sophistication of cyberattacks. Machine learning has gained great interest in securing data systems because it offers the capability of automatically detecting and responding to security threats in real time, which is crucial for maintaining the security of computer systems and protecting data from malicious attacks. This study concentrates on phishing attack detection systems, a prevalent cyber-threat. These systems assess the features of the incoming requests to identify whether they are malicious or not. Although the number of features is increasing in these systems, feature selection has become an essential pre-processing phase that identifies the most important features of a set of available features to prevent overfitting problems, improve model performance, reduce computational cost, and decrease training and execution time. Leveraging genetic algorithms, known for simulating natural selection to identify optimal solutions, we propose a novel feature selection method, based on genetic algorithms and locally optimized, that is applied to a URL-based phishing detection system with machine learning models. Our research demonstrates that the proposed technique offers a promising strategy for improving the performance of machine learning models.

We use technological developments in education and adapt technology to education. Educational computers, tablets, smart phones and other technological devices that integrate with these devices aim to increase the quality of education within the scope of computer-aided training. At the same time, with the development of the Internet, it has become much easier to access resources. The environments where the hardware and software technologies used in education are used together can be considered as intelligent classes. Video conferencing and live broadcasts used in the first smart classrooms have now turned into different applications. Intelligent classes that support learning with robotics, mobile learning, virtual reality and augmented reality applications, and different learning environments and materials are used today to improve the quality of education. The Individualized Smart Class (BAS) application proposed in this study is designed in two steps, hardware and software. It is envisaged to use laptop computers, microphones, headphones, tablets and virtual reality glasses in the classroom as hardware. At the same time the broadband of the classes will be ensured to have the internet. Even in this class, wireless network systems such as infrared, Bluetooth, Wi-Fi can be used. As a software, a class that can be included in the virtual classes with cloud architecture and can use the increased reality applications is considered. In addition, mobile applications that provide virtual reality will enrich the course materials. With this application, equality of opportunity will be provided for disadvantaged students in education and training, quality of education will be improved and it will be beneficial for the development of our country in this context.

The feedback of college students by surveys is rather difficult. Twitter provides a large data set for student sentiment analysis. In this article, both lexical and classification based approaches have been tested. Six different classification algorithms, Naive Bayes, SVM, Bayesian Network, C4.5 Decision Tree, Bagging and Random Forest, were applied to the classification based approach and the results were compared with each other.. 10 fold cross validation was used during the test phase. Similar results have been observed for Twitter sentiment analysis when looking at the results. Within the scope of this study, Based on a Lexicon based approach the happiness rates were compared for six universities, three of which are state universities and three of which are foundation universities in Istanbul. When the results are compared, it gives unhappy results for a university while giving the happy results for five universities. This study for university sentiment classification will give more accurate results with more data. In addition, it is thought that more accurate results will be obtained by increasing the number of words in the dictionary. A work to be done in this regard will increase the accuracy of the current work.

Today, with the widespread use of the Internet, users are sharing more with online platforms such as social networks and websites. They also pass on their feedbacks on their products and services to both product/service providers and other users via these platforms. This process needs to be automated as the amount of these feedbacks reaches a size that cannot be examined individually by humans. A large number of dictionary-based sentiment analysis studies have been conducted on English texts in the literature. However, the number of studies done in this area for Turkish is very low. Accordingly, in this research, a sentiment analysis study for Turkish customer comments is presented. Within the scope of the study, an adjective dictionary consisting of 9,822 adjectives was scored by the authors in the range [-2, +2]. By using this dictionary, 3-class (positive, negative and neutral) sentiment analysis study was conducted on 2,100 book reviews collected from an online book sales site. Besides the adjective dictionary, the success rate of the study has been increased by developing Turkish language-specific rules. As a result, 61.19% accuracy and 0.5935 F-measurement value were obtained.