Detection Of Fake Online Reviews Using Semi-Supervised And Supervised Learning
Main Article Content
Abstract
In the era of booming social media and e-commerce, the authenticity of online reviews has become a critical factor influencing consumer decision-making. This study presents an innovative approach to combating the proliferation of fake online reviews, combining semi-supervised and supervised learning techniques. Our method involves extracting pertinent features from review text and employing them to train a supervised classifier, augmented by a semi-supervised algorithm designed to identify and eliminate potentially fraudulent reviews. We leverage a comprehensive dataset comprising genuine and fake reviews, and our results demonstrate the superior performance of our proposed method compared to existing techniques. It achieves a high accuracy rate in identifying fake reviews while effectively minimizing false positives. in parallel, the increasing importance of social media monitoring highlights the significance of analyzing social data to gain insights into consumer behaviour. To this end, we conduct sentiment analysis on Twitter tweets, focusing on user reviews of movies. This paper introduces a novel combined dictionary that incorporates social media keywords and online review data. Additionally, we uncover hidden relationship patterns among these keywords, shedding light on the intricate dynamics of online sentiment expression.
As online shopping continues to gain popularity, online reviews have emerged as pivotal factors in e-commerce decision-making. A substantial portion of consumers routinely consult product and store reviews before making purchasing decisions. Unfortunately, this heightened reliance on reviews has given rise to a surge in fake, spam, or poorly constructed reviews that can mislead consumers. Therefore, the need to distinguish genuine from fraudulent reviews is more critical than ever.
Our study addresses this issue by leveraging machine learning techniques, including supervised and semi-supervised learning, to accurately detect fake reviews. Recognizing that obtaining labelled data can be challenging, we emphasize developing models requiring minimal labelled data. Our approach ensures efficiency, providing rapid results while maintaining high accuracy.
In this paper, we explore various classification algorithms, including support vector machines (SVM), random forests (RF), and deep neural networks, to discern the authenticity of online reviews. By combining the power of these algorithms and the use of unlabeled data through semi-supervised learning, we strive to enhance the credibility of online reviews and promote informed decision-making in the realms of e-commerce and social media.
Article Details
References
B. Erçahin, Ö. Aktaş, D. Kilinç, and C. Akyol, ‘‘Twitter fake account detection,’’ in Proc. Int. Conf. Comput. Sci. Eng. (UBMK), Oct. 2017, pp. 388–392.
F. Benevenuto, G. Magno, T. Rodrigues, and V. Almeida, ‘‘Detecting spammers on Twitter,’’ in Proc. Collaboration, Electron. Messaging, Anti- Abuse Spam Conf. (CEAS), vol. 6, Jul. 2010, p.
S. Gharge, and M. Chavan, ‘‘An integrated approach for malicious tweets detection using NLP,’’ in Proc. Int. Conf. Inventive Commun. Comput. Technol. (ICICCT), Mar. 2017, pp. 435–438.
T. Wu, S. Wen, Y. Xiang, and W. Zhou, ‘‘Twitter spam detection: Sur- vey of new approaches and comparative study,’’ Comput. Secur., vol. 76, pp. 265–284, Jul. 2018.
S. J. Soman, ‘‘A survey on behaviors exhibited by spammers in popular social media networks,’’ in Proc. Int. Conf. Circuit, Power Comput. Tech- nol. (ICCPCT), Mar. 2016, pp. 1–6
A. Gupta, H. Lamba, and P. Kumaraguru, ‘‘1.00 per RT #BostonMarathon # prayforboston: Analyzing fake content on Twitter,’’ in Proc. eCrime Researchers Summit (eCRS), 2013, pp. 1–12.
F. Concone, A. De Paola, G. Lo Re, and M. Morana, ‘‘Twitter analysis for real-time malware discovery,’’ in Proc. AEIT Int. Annu. Conf., Sep. 2017, pp. 1–6.
C. Chen, Y. Wang, J. Zhang, Y. Xiang, W. Zhou, and G. Min, ‘‘Statistical features-based realtime detection of drifted Twitter spam,’’ IEEE Trans. Inf. Forensics Security, vol. 12, no. 4, pp. 914–925, Apr. 2017.
C. Buntain and J. Golbeck, ‘‘Automatically identifying fake news in popu- lar Twitter threads,’’ in Proc. IEEE Int. Conf. Smart Cloud (SmartCloud), Nov. 2017, pp. 208–215.
S. Keretna, A. Hossny, and D. Creighton, ‘‘Recognising user identity in Twitter social networks via text mining,’’ in Proc. IEEE Int. Conf. Syst., Man, Cybern., Oct. 2013, pp. 3079–3082.
S. Sadiq, Y. Yan, A. Taylor, M.-L. Shyu, S.-C. Chen, and D. Feaster, ‘‘AAFA: Associative affinity factor analysis for bot detection and stance classification in Twitter,’’ in Proc. IEEE Int. Conf. Inf. Reuse Integr. (IRI), Aug. 2017, pp. 356–365.
M. U. S. Khan, M. Ali, A. Abbas, S. U. Khan, and A. Y. Zomaya, ‘‘Segregating spammers and unsolicited bloggers from genuine experts on Twitter,’’ IEEE Trans. Dependable Secure Comput., vol. 15, no. 4, pp. 551–560, Jul./Aug. 2018.