Text Plagiarism Detection Method Based On Path Patterns

摘要

This paper extends the forward method plagiarism detection in finding the percentage of similarity between documents. We have developed an algorithm to quantify the similarity based on path patterns, and the method employed is simple, as it involves only ordinary mathematics, thus simplifying application programming and speed up processing time. The method simply converts words into steps, which walks on a mesh, using a new proposed hash function. The hash function guarantees that the number of steps for each different word is unique and thus the walk pattern on a mesh is unique. Hence, a plagiarised version document will display a unique pattern on a mesh that is similar to the original document. This extended paper presents the algorithm in detail, and results are compared with an available online plagiarism detection tool.