<![CDATA[Plagiarism detection is a complex task. In-text, it should be able to find fragments of a text that is suspected of being illegally plagiarized from other sources. Aligning the plagiarized passages of suspicious documents from the source document is an issue that was discussed a lot, of which we can measure the percentage of the plagiarized text. This research proposes a semantic approach of text (fragments in documents) alignment between source and suspicious documents, using Jackard similarity method. Experimental results on the PAN competition for plagiarism detection competition, yielding average of 66.9% detection scores, increased more than twice if compared to the baseline method provided by the organizer, which is 28,4%. This approach is potential as a starting point to find offset match and length of plagiarized text in a plagiarism detection system. ]]>
Copyrights © 2021
