Abstract: In recent years, a substantial growth has been experienced in the mobile phone market. A cumulative of 432.1 million mobile gadgets has delivered in the second quarter of 2013 with an increment of 6.0% year over year. As the acquisition of cellphone gadgets has become common, Short Message Service (SMS) has developed into a multi-billion-dollar business. A rush in the quantity of unwanted business notices sent to cell phones utilizing text messages has additionally expanded due to the increased popularity of mobile platforms. This rise attracted attackers, which have resulted in SMS Spam problem. This study presents model for SMS spam filtering classification using Deep Machine Learning Techniques. The system uses the deep machine learning model(MLPNM) in tensorflow and keras framework to classify SMS Message dataset containing 5574 messages. The dataset was read from directory using the pandas.read_csv function. The dataset was cleaned to make sure there are no null values present. The Deep learning model was built with a total of three dense layer with takes in 8672 inputs and 1 output, a batch size (batch size equals the total dataset thus making the iteration and epochs values equivalents) of 32 and epoch value of 50. This trained model was saved and exported into web for easy access and testing with the help of python flask, so that users make various input SMS message. Bootstrap framework (HTML and CSS) was use to design the Front End, while for the Backend, python programming language was use. The results of the test showed accuracy of 99.82% of all input message classified as either ham(legimate) or spam to verify if it’s actually a Spam SMS message or a Ham (Legitimate) SMS spam messages.

Keyword: SMS, Spam, Deep Learning, Tensorflow, Keras.

PDF | DOI: 10.17148/IJARCCE.2021.10403

Open chat
Chat with IJARCCE