Detection of Social Media Hashtag Hijacking Using Dictionary-based and Machine Learning Methods

Nowadays, hashtags are widely utilized on all social media platforms since they deliver numerous benefits, particularly for corporations aiming to reach a larger audience. However, hashtag exploitation has resulted in the problem of hashtag hijacking, which is a type of cyber content threat that anyone or any organization can carry out. As a result, this research presents a framework for detecting social media hashtag hijacking through machine learning algorithms. This paper aims to identify methods to classify relevant and irrelevant hashtags to their contents. This paper demonstrates the unsupervised machine learning method, namely the dictionary-based approach, to classify the relevance of hashtags with the content of tweets on an unlabeled dataset, and also the implementation of supervised machine learning methods, including the Support Vector Machine (SVM), Naive Bayes classifier, and Decision Tree algorithms, to classify the relevance of hashtags used with their contents and compare the machine's performances on labeled datasets. Our results showed that the Support Vector Machine (SVM) performs the best in classifying the relevance of hashtags with an accuracy of 93.36%, an F1 score of 96.19% and ROC-AVC score of 97.22 %. The findings of the study present an automated detection framework for hashtag hijacking that can overcome the limitations of previous studies and adapt to external threats with high performance over time.