Classifying Suspicious Content on Social Media Networks
Noor Alwan Ghanem, Haider M. Habeeb · 2021
With the emergence and expansion of web technologies, the web now contains a huge amount of data and information for Internet users, in addition to a large amount of new data being generated at every moment. The Internet, with its various platforms, has evolved to be a forum for learning through these platforms, exchanging information, and exchanging opinions and ideas. For example, Twitter and other social networks are rapidly gaining popularity because they encourage users to communicate and convey their opinions on a variety of topics, engage in discussions with different cultures, and send messages around the world. In the areas of sentiment about Twitter information, a lot of studies have been done. This study focuses on the use of sentiment analysis technology on tweets generated by Twitter which is useful for analyzing details in tweets where opinions are highly disorganized, heterogeneous, and either negative, positive, or neutral in some cases. In this paper, the mentioned technique (sentiment analysis) was used to elicit opinions on two suspicious or non-suspicious measures as well as categorize these tweets using machine learning and a lexicon-based approach, along with rating scales. Using different machine learning algorithms such as Naive Bayes and Random Forest Classifier (RFC), we achieved a classification accuracy of 88.07 and 92.61 for NB and RFC. The accuracy of NB and RFC is 88.07 and 92.61, respectively, which is very good when compared to the recent research mentioned below.