<![CDATA[Hate speech (HS) is defined as speech that conveys hateful meaning and intent. In contemporary times, the prevalence of hate speech has surged in the virtual realm, particularly on social media platforms. Among these platforms, Twitter, now renamed X, stands out as one of the most widely used and a significant medium for the dissemination of hate speech. Hate speech can be categorized into various levels of severity, including HS_Weak, HS_Moderate, and HS_Strong. This study utilizes a dataset comprising 13,169 tweets from the social media application X from Indonesia users in 2023 to investigate hate speech detection. The research employs a novel hybrid approach that integrates image input with five preprocessing techniques: data cleaning, case folding, tokenization, stop-words removal, and stemming. Following preprocessing, the study applies Natural Language Processing (NLP) techniques in conjunction with Naïve Bayes classification. The combination of these NLP methods proves to be highly effective for the classification of text data. The key findings of this research demonstrate that the hybrid method significantly enhances hate speech detection accuracy. The evaluation of the classification model, based on training and validation, reveals an accuracy rate of 80%, a precision value of 85%, a recall value of 75%, and an F1-score of 80%. These results indicate substantial improvement over previous research outcomes. The findings suggest that the hybrid method is robust and effective for hate speech detection on social media platforms. Future research should explore the comparison of this hybrid approach with other classification methods to further validate its efficacy and potential applications in various domains of text classification.]]>
