Source Code Plagiarism Detection via Design Pattern Analysis

The patent segments the plagiarism detection process into multiple stages: first extracting design patterns from source code, then comparing pattern structures rather than raw code text. This segmentation allows the system to handle split and modified code by focusing on the underlying pattern architecture rather than surface-level syntax, thereby resolving the contradiction between simple comparison and accurate detection.

The patent introduces design patterns as an intermediary layer between the source code and the comparison process. Instead of directly comparing source code files, the system extracts and compares design pattern instances, which serve as a mediator that reveals the structural similarity even when code is split across files or modified in format. This intermediary approach maintains detection accuracy while preserving system simplicity.

The patent applies preliminary action by pre-defining a database of design patterns and their structural characteristics before the plagiarism detection process. This pre-prepared knowledge base allows the system to quickly match and compare pattern structures without performing complex real-time analysis, thereby reducing system complexity while maintaining high detection accuracy for split and modified code.

The patent changes the detection parameters from text-based string matching to structure-based pattern matching. By transforming the comparison criterion from surface-level code text to underlying design pattern architecture, the system achieves higher accuracy for modified and split code without proportionally increasing complexity, as the pattern matching operates on standardized structural representations.