Learning-Based Static File Extraction for Software Authenticity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing software authenticity determination methods, such as static analysis, dynamic analysis, and selection using snapshots, are limited in accuracy and scope, leading to potential misidentification of files and reduced reliability in detecting unauthorized software falsification.
Innovation Solution
A learning model is generated to distinguish between static and dynamic files by analyzing file features, allowing for accurate identification and extraction of static files as determination targets, using a support vector machine for classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods (static analysis, dynamic analysis, snapshot) are used to select static files, then file selection can be performed, but accuracy of software authenticity determination deteriorates due to misidentification of files
Solution Approach 1:
The patent replaces conventional mechanical/file-system-based methods (static analysis relying on meta information, dynamic analysis using system call monitoring, snapshot comparison) with a machine learning-based classification system. The learning model automatically identifies static files by analyzing file features and patterns, substituting the manual and rule-based approaches with an intelligent system that achieves higher accuracy in distinguishing static from dynamic files, thereby improving both measurement precision and reliability of software authenticity determination
Solution Approach 2:
The patent changes the parameters used for file classification from conventional metadata and time-based monitoring to multiple file features including file path, file size, creation time, modification time, and other attributes. By transforming these parameters into a feature vector and applying machine learning classification, the system achieves more accurate identification of static files compared to traditional single-parameter or rule-based methods
2Measurement precision
If dynamic analysis with sufficient monitoring time is used, then accuracy of static file selection is improved, but files that do not change accidentally or files with long change cycles may be incorrectly included
Solution Approach 1:
The patent performs preliminary classification of files as static or dynamic before the actual authenticity determination process. By using the learning model to pre-identify static files based on their inherent features and patterns, the system avoids the problem of dynamic analysis where files may appear static during monitoring periods but change later. This preliminary action ensures that only truly static files are selected for authenticity verification, improving reliability without requiring extended monitoring periods
Data Source
AI summary
An information processing device having a memory and a processor. The processor is coupled to the memory and programmed to execute a process including learning each feature of a dynamic file and a static file to generate a learning model; acquiring a predetermined file group at a reference point of time from an external device that uses the predetermined file group; and extracting a determination target file, which is the static file, from the predetermined file group on the basis of the learning model.