Similarity Overlap Metric and Greedy String Tiling at PAN 2012: Plagiarism Detection

This paper reports the best performed approach followed for the candidate document retrieval task and the approach used for the detailed comparison task of the Plagiarism detection track in PAN 2012. The aim of the participation was to understand a few of the computer-assisted approaches used for plagiarism detection. The plagiarism detection is dependent on two broad tasks, (1) the candidate document retrieval task and (2) the detailed comparison task. The N-gram similarity overlap metric was used for candidate document retrieval task and the greedy string tiling algorithm for detailed comparison task. The evaluation results suggested that the approach used for the candidate document retrieval task was highly competitive, but the approach used for detailed comparison task need much more improvement.