AI-Based Plagiarism Detection System

Abstract

The AI-Based Plagiarism Detection System, named PlagiChek, is an intelligent solution designed to overcome the limitations of traditional plagiarism detection tools that rely primarily on keyword and string matching. Conventional techniques are ineffectual against AI-generated or rephrased plagiarism because they frequently fail to detect paraphrased or semantically restructured text. The proposed approach utilizes semantic similarity analysis and natural language processing (NLP) to more accurately identify conceptually comparable information, addressing this gap. The system tokenizes, preprocesses, and creates transformer-based sentence embeddings to process and compare two text documents. It was developed using Python, Streamlit, Sentence Transformers (MiniLM), NLTK, NumPy, and ReportLab. These embeddings enable the model to calculate the semantic similarity of sentences, resulting in a precise similarity percentage displayed via an interactive Streamlit interface. Furthermore, the system generates color-coded similarity reports—green for unique, orange for moderate similarity, and red for extreme similarity—as well as downloadable PDF summaries for in-depth research. The goal is to create an advanced plagiarism detection technology that can identify reworded and contextually identical content in addition to word matching. This approach has useful applications in academic and research settings, supporting institutions, instructors, and students in ensuring academic integrity and originality in research papers, theses, and assignments.