Bullying has moved online as a result of the technological revolution, which was previously limited to physical boundaries. One type of cyber bullying is ridicule or insult. According to the report the cyber bullying on social media is getting worse. Insulting words change over time, and the same word can mean different things depending on the situation. A comment cannot be considered bullying simply because it contains such a word. Therefore, simple keyword spotting methods are insufficient for labelling comments. Lexical databases like Word Net, which provide synonyms and homonyms for words, have been utilized in other languages to address this issue. It is difficult to identify a word as bullying because there is no English-language lexical database. As a result, the proposed work solved the problem by following the rules. Outliers were removed from the collection of tweets containing profane language, and the remaining tweets were pre-processed. Five feature extraction rules were used to find insults in the text. The Support Vector Machine (SVM), K-nearest neighbor (KNN), and Naive Bayes algorithms were then utilized. With F1-score of 91 percent, the findings demonstrate that SVM along with an RBF kernel performs better. The fact that this research focuses on English-language cyberbully detection is novel and has not been done before.