Development of an Algorithm for Plagiarism Detection

For many years, plagiarism has remained a serious problem in Higher Learning Institutions (HLIs). Despite having adverse effects on the quality of education, plagiarism has been rapidly increasing in HLIs across the globe. Many researchers agree that the ease of access to information over the internet has made plagiarism a common occurrence in Tanzanian HLIs. In order to address this problem, many Tanzanian HLIs have enacted strict anti-plagiarism policies that require the use of software to detect plagiarism cases efficiently. Although many free and commercial plagiarism detection software exist, HLIs in Tanzania face numerous challenges in finding appropriate tools. The accuracy and effectiveness of freely available plagiarism detection software have been continuously questioned as they often provide inconsistent results. Moreover, commercial software that promise better performance have high annual subscription fees that are not easily affordable by HLIs in developing countries. This study aimed to address the need for affordable and reliable plagiarism detection tools in Tanzanian HLIs by developing an efficient algorithm for plagiarism detection. The study employed a systematic approach that involved different stakeholders in Tanzanian HLIs, including students, academic staff, and support staff. Questionnaires, unstructured interviews, and thorough literature analysis were used to identify the stakeholders’ needs and establish user requirements. The study proposed a plagiarism detection algorithm using a machine learning approach to information extraction, graph-based information retrieval, and semantic textual similarity methods. A web-based plagiarism detection system that implements the proposed algorithm was developed using open source technologies such as Symfony web framework, Neo4j graph database, MySQL database, and RabbitMQ. User Acceptance Testing (UAT) results concluded that the stakeholders positively accepted the developed algorithm. Furthermore, the developed web-based plagiarism detection system has received a Copyright Clearance Certificate from The Copyright Association of Tanzania (COSOTA), and The Committee of Vice Chancellors Principles and Provosts of Tanzania (CVCPT) has recommended the deployment of the system in Tanzanian HLIs.