Smishing Detector

Author: Antonio Artur de Holanda e Ayres de Moura

Smishing detector model trained and deployed on SageMaker and served via lambda functions. Detailed information is found in the project report.

Setup

Use the requirements file to install necessary libraries.

pip install -r requirements.txt

Dataset

To download the dataset, execute the following commands in order

wget https://md-datasets-cache-zipfiles-prod.s3.eu-west-1.amazonaws.com/f45bkkt8pr-1.zip
unzip f45bkkt8pr-1.zip
unzip "SMS PHISHING DATASET FOR MACHINE LEARNING AND PATTERN RECOGNITION/Dataset_5971.zip"

Notebooks

EDA:

Contains the Exploratory data analysis code.

SciKit-Learn

Contains the experiments with scikit-learn models.

BERT

Contains the experiments with BERT model.

Training and deployment

To train and deploy the model simply upload the contents of ./code/SageMaker/training to a notebook instance and execute all cells of "Training Notebook.ipynb".

Lambda Function

• Create a lambda function using ./code/SageMaker/lambda-function/lambdafunction.py

obs: Change the endpoint name to the one you deployed at the training and deployment step