Abstract:
Nowadays,the number of mobile users is always increasing. SMS stands for "short
messaging service," which lets the users send and receive short text messages on regular
phones as well as smartphones.
The quantity of SMS texts increased significantly as a result. Additionally, Spam, or
unsolicited messages, became more prevalent. Spammers send unsolicited emails with
the intention of gaining business or money through things like buying lottery tickets,
breaking into new markets, or disclosing credit card details. This directly leads to more
attention being paid to sorting via spam. There exist numerous content based ML
(machine learning) strategies that have been shown to be successful in removing spam
from emails. Researchers of today have classified text messages as spam or ham by using
certain stylistic elements. The actual existence of all well-known terms,phrases,
acronyms, and idioms can have a great significant impact on SMS spam detection. This
study compares various classification methods using various datasets gathered from
earlier research projects.This paper proposed a powerful solution based on machine
learning classification techniques. this paper developed and tested and evaluated this
strategy utilizing five learning algorithm : Naive bayes with 97% accuracy, k-Nearest
Neighbors Algorithm with 89% accuracy, Decision tree learning algorithm with 96%
accuracy, SVM algorithm with 94% accuracy, Random Forest algorithm with 95%
accuracy level. The experimental data demonstrated that all of the proposed methods
provide very high levels of accuracy for recognizing these data sets but Naive Bayes tops
them all with 97% accuracy.
