Yazar "Akkol, Ekin" seçeneğine göre listele

Listeleniyor 1 - 4 / 4
  • Küçük Resim Yok
    Öğe
    CREATING A COMPREHENSIVE DATA SET FOR DECEPTION DETECTION STUDIES IN TURKISH TEXTS
    (Suat TEKER, 2024) Akkol, Ekin; Gökşen, Yılmaz
    Purpose- Deception detection has gained increasing importance with the widespread use of digital communication and online platforms. While numerous studies have been conducted on deception detection in various languages, a significant gap remains in the availability of a Turkish-language dataset for detecting deceptive reviews. This study addresses this gap by creating a comprehensive dataset specifically for deception detection in Turkish hotel reviews, including real, fake, and AI-generated comments. The dataset aims to facilitate research on deception detection, enhance the reliability of user-generated content, and contribute to the development of automated methods for identifying deceptive texts. Methodology- The study included a dataset of 5,013 Turkish hotel reviews, including real reviews from Tripadvisor, fake reviews generated by humans, and fake reviews generated by AI using the OpenAI GPT API. The collected dataset underwent extensive preprocessing to ensure quality and reliability, including data cleaning, filtering criteria, and balancing the distribution of real and fake comments. Descriptive and statistical analyses were performed to identify linguistic patterns and structural differences across these three categories. Specifically, linguistic features such as comment length, complexity, readability (measured using the Gunning Fog Index), and pronoun usage were examined. Findings- Real comments are longer and more detailed than fake and AI-generated comments, while fake comments are simpler and clearer, which supports deception detection studies in other languages. AI-generated comments frequently use the pronoun ‘we’, while fake comments tend to mimic personal experience with the pronoun ‘I’. In addition, the pronoun usage in real comments is more balanced and shows an authentic language structure. Conclusion- This study makes important contributions for fake comment detection by providing the first large-scale Turkish deception detection dataset. The findings can help businesses improve the credibility of online comments. Future work could focus on machine learning applications and comparisons with different languages.
  • Küçük Resim
    Öğe
    A framework for sustainable and data-driven smart campus
    (Scitepress, 2020) Köstepen, Zeynep Nur; Akkol, Ekin; Doğan, Onur; Bitim, Semih; Hızıroğlu, Abdülkadir
    As small cities, university campuses contain many opportunities for smart city applications to increase service quality and use of public resources efficiency. Enabling technologies for Industry 4.0 play an important role in the goal of building a smart campus. The study contributes to the digital transformation process of.Izmir Bakircay University which is a newly established university in Turkey. The aim of the study is to plan a road map for establishing a smart and sustainable campus. A framework including an architectural structure and the application process, for the development of a smart campus have been revealed in the study. The system application is designed to be 3 stages. The system, which is planned to be built on the existing information systems of the university, includes data collection from sensors and data processing to support the management processes. The proposed framework expects to support some value-added operations such as increasing personnel productivity, increasing the quality of classroom training, reducing energy consumption, accelerating interpersonal communication and finding the fastest solution to the problems on campus. Therefore, not only a smart campus but also a system is designed for sustainability and maximum benefit from the facilities.
  • Küçük Resim Yok
    Öğe
    In Text Classification, Bitcoin Prices and Analysis of Expectations in Social Media with Artificial Neural Networks
    (Burdur Mehmet Akif Ersoy University, 2020) Çılgın, Cihan; Ünal, Ceyda; Alıcı, Serkan; Akkol, Ekin; Gökşen, Yılmaz
    In recent years, Web 2.0 services such as blogs, tweets, forums, emails have been widely used as communication channel. Also, social media; it is considered to be the easiest and most up-to-date way to both share information and express opinions such as requests, complaints, and wishes. As in many fields, the effect of social media on Bitcoin prices has been addressed in the last few years. Bitcoin is an investment tool that has been underlined for years, and is increasing in popularity day by day. Bitcoin, an electronic currency system that is decentralized, states a radical change in financial systems that has attracted many users. In this study, interaction of social media with Bitcoin price was revealed, particularly based on tweets obtained from Twitter channel. For this purpose, various analyses were carried out by using classification algorithms in machine learning methods over a total of 2,819,784 tweets posted by Twitter users between 06.10.2018-19.05.2019. When the findings were evaluated, Artificial Neural Networks with the highest accuracy rate of 90% was used in text classification. In addition, bilateral correlations were made with Bitcoin prices and classified positive / negative tweet rates. The correlation coefficient of 0.681 was found to be positively correlated with higher than moderate strength.
  • Küçük Resim Yok
    Öğe
    Understanding patient activity patterns in smart homes with process mining
    (Springer International Publishing Ag, 2022) Doğan, Onur; Akkol, Ekin; Oluçoğlu, Müge
    Especially in people over 50 years of age, sedentary lifestyle can cause muscle loss called sarcopenia. Inactivity causes undesirable outcomes such as excessive weight gain and muscle loss. Weight gain can lead to a variety of problems, including deteriorating of the musculoskeletal system, joint problems, and sleep problems. In order to provide better service, it can be beneficial to understand human behavior in terms of health services. Process mining, which can be considered a part of knowledge graphs, is a crucial methodology for process improvement since it offers a model of the process that can be analyzed and optimized. This study uses process mining approaches to examine data from three patient that were collected using indoor location sensors, allowing the collection of flows of human behavior in the home. The analyses indicated how much time was spent by the patients of the house in each room during the day as well as how frequently they occurred. The movement of patients from room to room was observed daily and subjected to a variety of analyses. With the help of user pathways, lengths of stay in the rooms, and frequency of presence, it has been possible to reveal the details of daily human behavior. Inferences about the habits of the participants were revealed day by day.