The Use of Information Retrieval in Student Academic Document Plagiarism Detection System

Abstract

The practice of plagiarism can threaten the credibility of academic documents such as final assignments, theses, theses, and dissertations. Universitas Muhammadiyah Ponorogo (UMPO) is one of the best Muhammadiyah Universities in Indonesia. All academic documents at UMPO are well stored. However, plagiarism is only found in the Turnitin database and in the title of academic documents. This research aims to pre-screen academic documents in the internal scope by creating a plagiarism application that is easy to use and cheaper in terms of cost than using Turnitin. The steps of making a plagiarism detection system in academic documents using information retrieval (IR) with waterfall development. The design process starts from system requirements analysis, design, implementation, and testing. This plagiarism detection system goes through several stages, namely tokenisation, stopword removal, stemming, and termweigthing. Tokenisation involves converting uppercase letters into lowercase letters and removing punctuation marks. Stopword removal removes noise, stemming converts words into basic forms, and termweigthing uses local and global weighting to calculate similarity with the Vector Space Model. Document similarity is calculated using cosine similarity, rejected documents are considered free of plagiarism. If there is no similarity, the document is accepted as a source document, while the rejected document is not stored in the database. Further research on the document plagiarism detection system that previously only used the txt file extension, in the future it can use the .doc or HTML extension so as to increase the effectiveness of the performance of academic document examination time.