The final, formatted version of the article will be published soon.
METHODS article
Front. Artif. Intell.
Sec. Machine Learning and Artificial Intelligence
Volume 7 - 2024 |
doi: 10.3389/frai.2024.1401126
This article is part of the Research Topic Disinformation Countermeasures and Artificial Intelligence View all articles
OLTW-TEC: Online Learning with Sliding Windows for Text Classifier Ensembles
Provisionally accepted- 1 West Ukrainian National University, Ternopil, Ukraine
- 2 Kharkiv National University of Radioelectronics, Ukraine, Ukraine
- 3 Lviv Polytechnic National University, Lviv, Ukraine
In the digital age, rapid dissemination of information has elevated the challenge of distinguishing between authentic news and disinformation. This challenge is particularly acute in regions experiencing geopolitical tensions, where information plays a pivotal role in shaping public perception and policy. The prevalence of disinformation in the Ukrainian-language information space, intensified by the hybrid war with russia, necessitates the development of sophisticated tools for its detection and mitigation. Our study introduces the "Online Learning with Sliding Windows for Text Classifier Ensembles" (OLTW-TEC) method, designed to address this urgent need. This research aims to develop and validate an advanced machine learning method capable of dynamically adapting to evolving disinformation tactics. The focus is on creating a highly accurate, flexible, and efficient system for detecting disinformation in Ukrainian-language texts. The OLTW-TEC method leverages an ensemble of classifiers combined with a sliding window technique to continuously update the model with the most recent data, enhancing its adaptability and accuracy over time. A unique dataset comprising both authentic and fake news items was used to evaluate the method's performance. Advanced metrics, including precision, recall, and F1-score, facilitated a comprehensive analysis of its effectiveness. The OLTW-TEC method demonstrated exceptional performance, achieving a classification accuracy of 93%. The integration of the sliding window technique with a classifier ensemble significantly contributed to the system's ability to accurately identify disinformation, making it a robust tool in the ongoing battle against fake news in the Ukrainian context. The application of the OLTW-TEC method highlights its potential as a versatile and effective solution for disinformation detection. Its adaptability to the specifics of the Ukrainian language and the dynamic nature of information warfare offers valuable insights into the development of similar tools for other languages and regions. OLTW-TEC represents a significant advancement in the detection of disinformation within the Ukrainian-language information space. Its development and successful implementation underscore the importance of innovative machine learning techniques in combating fake news, paving the way for further research and application in the field of digital information integrity.
Keywords: disinformation, fake news, Online Learning, Classifier ensembles, machine learning
Received: 14 Mar 2024; Accepted: 30 Aug 2024.
Copyright: Ā© 2024 Lipianina-Honcharenko, Bodyanskiy, Kustra and IvasechkŠ¾. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
* Correspondence:
Khrystyna Lipianina-Honcharenko, West Ukrainian National University, Ternopil, Ukraine
Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.