Chaos enhanced intelligent optimization-based novel deception detection system

Abstract

Today, with the developing technology, access methods to information have also changed. Internet blogs and news sites, social media, etc. replace the traditional information access tools such as TV, radio, newspaper, and magazines. Cheaper and faster access than traditional methods and easy access from anywhere where the internet is available are the main factors in changing the access method to information. Besides, information spreads rapidly on the internet without proving the accuracy. There may be many reasons for the distribution of misinformation in this way, such as commercial, political, and economic. Information with fake content and deception purposes negatively affects both the person and the society. It is extremely important to detect deceptive information in textual data and researchers continue to propose new models in order to obtain efficient results in terms of many metrics. This paper proposes a new approach for deception detection problems. The deception detection problem was considered an optimization problem for the first time. Optics Inspired Optimization (OIO), Grey Wolf Optimization (GWO), and Chaos Based Optics Inspired Optimization (CBOIOs) are adapted and modeled for the first time in the deception detection problem. Deception detection models proposed in this article include data preprocessing, adapting optimization methods, and testing stages. An experimental evaluation and comparison of the proposed model and seven supervised machine learning algorithms are performed within two different data sets depending on four evaluation metrics (Accuracy, Recall, Precision, and F-Measure). Results show that CBOIOs are more effective than OIO, GWO, and other machine learning algorithms for this problem.