Abstract:
Social media has become very popular in this technology-driven era, and most people
social media for business and many other purposes. The Internet has made this
possible to connect with the globe. Where this technology is used tor a positive sense,
people used it for negative. Some people use these platforms for a bully, threats,
and even terrorism discussion. The purpose ofthe proposed system is to monitor and
identify the suspicious communications on Twitter by identifying the suspicious words
in chats of users. To get effective results, we used machine learning algorithms to train
our model. Machine learning is a spectacularly powerful tool to make predictions or
forecasting based on the bulk amount of data. Data in text format should be cleaned to
reduce the noise within. We have done pre-processing in which the text present in the
post or message will be filtered, cleaned, and extracted.
use
some
The root words and unnecessary data cleanout, which increase processing time
and reduce cost. Then, in the next phase, feature selection and extraction of the text
normalization, tokenization, stemming. After the extraction of completed using case
features present in the text, feature weighting is done using techniques like TF/IDF.
have used four different algorithms for better accuracy. The For identification, we
algorithm we applied is Multinomial Naive Bayes, Decision Tree, Support Vector
Machine, and Adaboost. We have compared their result and then select the best one.
the results captured through software called Jupiter notebook using python All
language. All the details of our work are mentioned in this report.
