A Hybrid Graph Neural Network Approach for Detecting PHP Vulnerabilities

TL;DR

This paper introduces DeepTective, a hybrid deep learning model combining Gated Recurrent Units and Graph Convolutional Networks to effectively detect PHP code vulnerabilities like SQLi, XSS, and OSCI, outperforming existing methods.

Contribution

The paper proposes a novel hybrid deep learning approach that leverages syntactic and semantic features for PHP vulnerability detection, validated on synthetic and real-world datasets.

Findings

Achieves near-perfect classification on synthetic data

Attains an F1 score of 88.12% on real-world data

Discovers 4 new vulnerabilities in WordPress plugins

Abstract

This paper presents DeepTective, a deep learning approach to detect vulnerabilities in PHP source code. Our approach implements a novel hybrid technique that combines Gated Recurrent Units and Graph Convolutional Networks to detect SQLi, XSS and OSCI vulnerabilities leveraging both syntactic and semantic information. We evaluate DeepTective and compare it to the state of the art on an established synthetic dataset and on a novel real-world dataset collected from GitHub. Experimental results show that DeepTective achieves near perfect classification on the synthetic dataset, and an F1 score of 88.12% on the realistic dataset, outperforming related approaches. We validate DeepTective in the wild by discovering 4 novel vulnerabilities in established WordPress plugins.