Measuring Text Similarity in Engineering Students’ Technical Reports: A Cosine Similarity-Based Approach for Plagiarism Detection

Abstract

In engineering education, detecting plagiarism in technical reports is a significant challenge for instructors, particularly in large cohorts where manual comparison is impractical. This paper presents an automated, open-source tool designed to measure text similarity in students’ technical reports using Term Frequency-Inverse Document Frequency (TF-IDF) vectorization and cosine similarity. The tool generates a similarity matrix, identifies documents exceeding a customizable 65% threshold, and provides detailed outputs for instructor review. By facilitating similarity detection, the tool promotes independent technical writing, a critical skill for future engineers. Evaluation on a dataset of engineering reports demonstrates its effectiveness, with results displayed in a user-friendly graphical user interface (GUI) and text reports. The tool’s scalability and adaptability make it suitable for technical disciplines, including geoinformatics. Limitations, such as handling semantic paraphrasing, and future enhancements, such as integrating optical character recognition (OCR), are discussed.