Stepped Context Analysis for Plaintext Password Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for detecting and securing plaintext passwords on storage devices suffer from high false negatives, high false positives, and high computational costs due to the idiosyncratic nature of password creation and the complexity of free-flowing text without syntax or semantics, leading to overfitting and inefficiencies in existing techniques.
Innovation Solution
A multi-stepped context analysis approach is employed, utilizing file-based, content-based, and proximity-based contexts, combined with machine learning models, to identify potential passwords, followed by a kernel-based protection mechanism to authenticate users before accessing sensitive files.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional password detection techniques are used to scan and analyze files for plaintext passwords, then password detection capability is provided, but high false negatives and high computational cost occur due to the idiosyncratic nature of password creation and the large number of files to scan
Solution Approach 1:
The patent segments the password detection process into two distinct passes: a first pass that performs both password scanning and context scanning simultaneously to quickly identify potential passwords and their surrounding contexts, and a second pass that performs detailed analysis only on files where the first pass found potential passwords or promising contexts. This segmentation reduces computational overhead by avoiding full analysis of all files while maintaining detection accuracy through targeted second-pass analysis.
Solution Approach 2:
The patent applies preliminary action by performing context scanning in the first pass to identify files with contextual indicators of passwords (such as proximity to keywords like 'password', 'secret', or specific file patterns) before conducting full password detection analysis. This preliminary context assessment allows the system to prioritize which files require intensive second-pass analysis, reducing overall computational cost while improving detection reliability.
2Reliability
If conventional password detection techniques are used to scan files for plaintext passwords, then password detection is performed, but high false positives occur due to incompleteness of breached datasets and absence of context
Solution Approach 1:
The patent introduces context information as an intermediary element that mediates between the detected password candidate and the final determination of whether it is a true password. By analyzing surrounding text contexts, file metadata, and proximity to known password-related patterns, the system uses context as a mediator to verify password candidates, significantly reducing false positives while maintaining detection accuracy.
Solution Approach 2:
The patent changes the detection parameters dynamically based on context analysis. Instead of using fixed thresholds for password detection, the system adjusts detection sensitivity and analysis depth based on contextual indicators found in the first pass. Files with strong contextual evidence trigger more rigorous second-pass analysis, while files with weak context are handled with adjusted parameters, improving precision by adapting measurement criteria to each file's specific context.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
The disclosure relates generally to methods and systems for building rich context for effective password detection in plaintext. Detecting and securing plaintext passwords on hard-disk or storage device is difficult as humans generate passwords in a variety of idiosyncratic ways which results in high false negatives and involves a high computation cost. In the present disclosure, a stepped context analysis is performed which applies different context discovery strategies in sequential manner. In scenarios, where a potential password is unable to find in a file using simple detection method, but if the context likelihood of the file is higher than the configured threshold then the file is searched again with more detailed techniques for detecting presence of a potential password. This selective second pass for a few files helps in reducing the false negatives while balancing the proposed solution's performance.