Inspiration
Phishing attacks are social engineering attacks, that are hard to analyze and predict, as they take advantage of human-computer interaction. Hence, an in-depth study of the behavioural model allows us to assess potential attackers in advance.
What it does
The project builds a model that identifies attackers based on the behaviours on the graph network. It takes advantage of the attacker's need for a widespread, hit-and-trial approach.
How we built it
The model was built using PyTorch
Challenges we ran into
Some of the challenges I ran into were collecting or synthesising probable data.
Accomplishments that we're proud of
I am genuinely ecstatic about the variance analysis I did to create a new metric that allows for the study of the second layer of connections of a graph and utilizing that metric to predict the behaviour of the attacker.
What we learned
I had to adopt a new Graph Library (PyTorch-geometric) on the fly, and applying it was super fun and interesting.
What's next for Capturing Phishing Scams with GCN
The next step involves reverse-analyzing the content body, to associate attacks with certain attackers. This would help in tackling the issue of masked IP attacks.
Log in or sign up for Devpost to join the conversation.