Chrome extension permission risk assessment method based on dimensionality reduction driven confidence rule base

Abstract

A Chrome extension permission risk assessment method based on dimensionality reduction-driven confidence rule base is applicable to the field of browser extension security. This invention proposes the PermDR-BRB framework based on DR-BRB. This framework introduces multi-strategy dimensionality reduction, including feature selection, weighted combination, PCA, and Fast t-SNE, into the confidence rule base model. This not only effectively assesses Chrome extension permission risks but also solves the problem of combinatorial rule explosion in the confidence rule base. Experimental results show that PermDR-BRB achieves a classification accuracy of up to 99.00% and an MSE as low as 0.000555, demonstrating the effectiveness and accuracy of the PermDR-BRB framework.

Images