The prevalence of misleading information poses a significant challenge to efforts aimed at combating misinformation spread through social media, as such content can adversely affect public opinion and decision-making. Organizations that engage in the business of varied diversity face significant barriers to creating smart and sound mechanisms to detect misinformation with precision. This problem is a strategic dilemma that requires rigorous investigation and efforts to produce systematic answers and reverse the proliferation of fake information, as well as to enhance the trustworthiness of the online information infrastructure. In response to these challenges, there is an increasing demand for robust systems capable of identifying false information. The current paper presents a new way of research with the background of the previous works that use the TruthSeeker2023 dataset. The method combines neural network structures and natural language processing methods to be able to identify obscene cues of deceptive signals in the production of tweets. Multilayer perceptron topologies, or Deep Neural Network (DNN) topologies, are defined by small hidden layers with discrete activation functions. To convert the textual data to numeric attributes, sophisticated text vectorization models, Count Vectorizer and the Term Frequency-Inverse Document Frequency (TF-IDF), were used. The proposed methodology is superior to the highest accuracy rate of 96, which has been previously achieved; it has a high rate of 99. This paper elevates the levels of misinformation recognition and shows how intelligent systems can protect social media against misinformation, thus improving the credibility of the entire internet world.
nabaa kareem (Mon,) studied this question.