Negative Results of Fusing Code and Documentation for Learning to Accurately Identify Sensitive Source and Sink Methods An Application to the Android Framework for Data Leak Detection

TL;DR

This paper presents CoDoC, a deep learning-based tool that combines source code and documentation to accurately identify sensitive source and sink methods in Android, improving precision over previous models but still facing challenges in real-world scenarios.

Contribution

CoDoC introduces a novel approach using deep learning and combined code-documentation analysis to better identify privacy-related API methods in Android apps.

Findings

CoDoC achieves 91% precision, recall, and F1 score in cross-validation.

It outperforms the state-of-the-art SuSi on the same dataset.

Real-world performance of machine learning models for privacy detection remains limited.

Abstract

Apps on mobile phones manipulate all sorts of data, including sensitive data, leading to privacy-related concerns. Recent regulations like the European GDPR provide rules for the processing of personal and sensitive data, like that no such data may be leaked without the consent of the user. Researchers have proposed sophisticated approaches to track sensitive data within mobile apps, all of which rely on specific lists of sensitive source and sink API methods. The data flow analysis results greatly depend on these lists' quality. Previous approaches either used incomplete hand-written lists that quickly became outdated or relied on machine learning. The latter, however, leads to numerous false positives, as we show. This paper introduces CoDoC, a tool that aims to revive the machine-learning approach to precisely identify privacy-related source and sink API methods. In contrast to previous approaches, CoDoC uses deep learning techniques and combines the source code with the documentation of API methods. Firstly, we propose novel definitions that clarify the concepts of sensitive source and sink methods. Secondly, based on these definitions, we build a new ground truth of Android methods representing sensitive source, sink, and neither (i.e., no source or sink) methods that will be used to train our classifier. We evaluate CoDoC and show that, on our validation dataset, it achieves a precision, recall, and F1 score of 91% in 10-fold cross-validation, outperforming the state-of-the-art SuSi when used on the same dataset. However, similarly to existing tools, we show that in the wild, i.e., with unseen data, CoDoC performs poorly and generates many false positive results. Our findings, together with time-tested results of previous approaches, suggest that machine-learning models for abstract concepts such as privacy fail in practice despite good lab results.