In recent year, the number of users in social media has been increased multiple times when compared to the past. So most of them are aware about the current affairs in day today life. It is the medium for the user to express the opinion without facing any difficulty. At the same there will be a lot of bullying occurs in social media. The bullying like abuse word, aggressive text or posting some unwanted message. So the women feel unsecured in the society. Although a lot of techniques and methodology has been raised, but still the problem remain same. The major problem is the abuse word can be eliminated by the mean of report to the particular social media. In this methodology the unwanted message can be truncated in between the sender and receiver itself. So there other person cannot be affected in the cyberbullying. Also the unstructured data can be increased in the social media. So it leads to complex for analyzing the text. With the help of sentiment analysis, we can easily filter the unwanted text. Based on the criteria, the paper has been proposed with the problem statement as a main goal and developed a system in machine learning.