Yazar "Karabulut, Murat" seçeneğine göre listele

Listeleniyor 1 - 4 / 4
  • [ X ]
    Öğe
    A Smart Virtual Assistant Answering Questions about COVID-19
    (Institute of Electrical and Electronics Engineers Inc., 2020) Ugurlu, Yusuf; Karabulut, Murat; Mayda, Islam
    Numerous information and news put forward by different sources during the COVID-19 pandemic have caused many questions about the epidemic in people's minds. In this study, a smart virtual assistant was developed in order to eliminate this information pollution about COVID-19 and to give correct answers to people's questions with information obtained from reliable sources. Intelligent virtual assistants are question answering system software that enable people to get answers easily and quickly to their questions curious about. The use of these virtual assistants is becoming widespread all over the world in many fields, especially in education, e-commerce, tourism, finance and health. However, the number of question-answering system studies on Turkish is very few and insufficient. In fact, no question-answering study was found in the field of health that is presented for Turkish in the literature. The smart virtual assistant presented within the scope of this study is a first in this respect. This smart virtual assistant answers two different types of questions. The first of these is frequently asked information questions about COVID-19 among the people. The second is the questions about numerical data such as number of COVID-19 cases, number of deaths, number of tests, etc. of countries in the world. It has been observed that the developed virtual assistant can answer the vast majority of users' questions about COVID-19. It is thought that this smart virtual assistant could be useful in fighting the COVID-19 pandemic. © 2020 IEEE.
  • [ X ]
    Öğe
    Artificial Intelligence Supported Turkish University Virtual Assistant
    (Institute of Electrical and Electronics Engineers Inc., 2021) Ugurlu, Yusuf; Karabulut, Murat
    Today, it is known that processes such as artificial intelligence, data mining and data processing are gaining more and more value. There are many algorithms and learning models in artificial intelligence processes. It is important that these models are selected according to the data that is desired to get the meaning, and also the data is prepared according to the model to be used in training. It is observed that artificial intelligence makes analytical approaches and facilitates and accelerates different studies in different sectors with automated structures, thus providing a significant savings in terms of cost and labor. These structures are making a name for themselves in different businesses and sectors, from recommendation systems to health services. Intelligent virtual assistants are at the forefront of these automated structures. Intelligent virtual assistants are also known as chatbots. In this study, a smart virtual assistant has been developed that can answer the most frequently asked questions about the university, such as department, fee, academic staff, transportation, and scholarship opportunities. For the data set to be used in the study, 387 different questions were collected from the frequently asked questions on the official websites of 70 different universities in Turkey. In addition to these questions, the dataset includes 4 daily speaking questions and a total of 391 different questions. In addition to the existing questions, variations of the questions were produced to strengthen the training data, and a total of 1236 questions were prepared. The most important difference of this study from similar smart virtual assistant applications is that it is specific to the Turkish language. Thanks to the flexibility of the platform, the developed smart virtual assistant can serve university students in the preferred period both through the mobile application and the website. It is thought that the performance of the smart virtual assistant, which was revealed within the scope of the study, can be increased by enriching the data set in the future and it can provide one-to-one counseling support to the students throughout the education process. © 2021 IEEE.
  • [ X ]
    Öğe
    Development of Browser Extension for HTML Web Page Content Extraction
    (IEEE, 2020) Karabulut, Murat; Mayda, Islam
    As the amount of content on the websites increases, automatic content extraction from Web pages becomes more important. Although many studies have been done in the literature on this subject, a method that fully solves the problem has not been revealed due to the flexible structure of HTML. The performances of the methods that show success at certain rates also decrease over time with the changing and developing Web structure. In this study, a browser extension was developed to automatically download text content on Web pages. This developed extension provides an output with 100% recall rate by cleaning the text content on the Web page from all tags and codes with a parser that utilizes the Document Object Model (DOM) structure. This browser extension that operates independently from the language has been tested on different types of popular Web sites in Turkey and has been shown to work successfully.
  • [ X ]
    Öğe
    Web Based Algorithm Exercise and Assessment Management System for Computer Programming Students: AlgoBug
    (Institute of Electrical and Electronics Engineers Inc., 2020) Karabulut, Murat; Mayda, Islam
    Today, gaining computer programming skills and understanding the logic of algorithm has become important not only for those working in this field but also for those working in almost every profession. However, various difficulties are encountered in programming education for many reasons. In order for students to overcome these difficulties, they have to solve many sample programming questions. Although there are some English-based resources on exercise questions for those trying to learn programming, there is not yet a comprehensive and sufficient resource for Turkish. In order to fulfill this need, a web-based algorithm exercise and assessment management system with Turkish content was developed in this study. In the system called AlgoBug, users solve the practice questions of many programming languages in order from easy to difficult. Users' solutions are assessed by other users in the role of evaluator who are more advanced in that programming language, and detailed feedback is given to them. In addition to written feedback, solutions are scored between 1 and 10 points, and solutions scoring 7 or more are considered 'correct'. In order to carry out tasks such as introducing the system to new users in the system and guiding them in the programming learning process, the former users who have completed all the questions in a programming language are also given a consultancy role. With the AlgoBug system, people who are new to programming are provided with the opportunity to practice abundantly on the topics they have learned, while helping them to understand and correct their mistakes and shortcomings with the detailed feedback they receive on their solutions. While the users continue to improve themselves by solving the questions in a programming language, on the other hand, evaluating the solutions of the lower level users in the evaluator role enables them to further consolidate what they have learned with the 'learning by teaching' approach. It is thought that the AlgoBug system will contribute to reducing the difficulties encountered in programming education. © 2020 IEEE.