Source Code Correlation Filter for Plagiarism Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing software tools for comparing source code struggle to differentiate between relevant and irrelevant correlations, making it difficult to identify plagiarism amidst other sources of similarity such as third-party code, common algorithms, and common author styles.
Innovation Solution
A software tool that filters out irrelevant correlations by allowing users to discard specific identifiers, statements, comments, and instruction sequences from the correlation results, using a post-process filter file to refine the correlation database and adjust the correlation scores accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If source code correlation analysis is performed to detect plagiarism, then plagiarism detection capability is improved, but false positives increase due to irrelevant correlations from third-party code, common algorithms, and common author styles
Solution Approach 1:
The patent extracts and removes irrelevant correlation factors from the analysis. Users can specify which factors (identifiers, statements, comments, instruction sequences) to exclude, and the system filters these out from the correlation calculation, thereby eliminating false positives while maintaining genuine plagiarism detection capability
Solution Approach 2:
The patent changes the parameters of the correlation analysis by allowing dynamic adjustment of which code elements are considered. Users can modify the weightings and inclusions/exclusions of different factor types (identifiers, statements, comments, instruction sequences), transforming the correlation calculation to focus only on relevant similarities
2Reliability
If multiple correlation analysis runs are performed to filter irrelevant correlations, then detection accuracy is improved, but time consumption increases
Solution Approach 1:
The patent performs preliminary configuration where users define their filtering criteria before running the correlation analysis. By pre-specifying which factors to exclude and how to weight different elements, the system can execute a single optimized analysis run rather than requiring multiple sequential runs, thereby reducing time consumption while maintaining accuracy
3Productivity
If detailed correlation analysis is performed on all code elements, then comprehensive detection is improved, but complexity of analysis increases
Solution Approach 1:
The patent segments the code analysis into distinct factor types: identifiers, statements, comments, and instruction sequences. Each segment can be independently controlled and filtered, allowing users to simplify the analysis by excluding specific segments (e.g., comments or identifiers) while maintaining comprehensive detection capability across the remaining elements
Data Source
AI summary
A software tool allows a user to filter out the results produced by a software source code correlation program that is executed on sets of software source code files. The tool allows the user to discard elements of source code that are not relevant to the analysis without requiring the user to run the correlation program multiple times.