Hesti Ari Wardani, Imam Yuadi

Analisis Visual dan Machine Learning untuk Mengukur Validitas Dokumen Akademik Perpustakaan: Studi pada Data Turnitin

Introduction

Analisis visual dan machine learning untuk mengukur validitas dokumen akademik perpustakaan: studi pada data turnitin. Analisis visual & machine learning mendeteksi validitas dokumen akademik via data Turnitin. Pahami distribusi skor, pola validitas, & tingkatkan evaluasi akademik.

Abstract

The issue of plagiarism in academic works is a major concern in higher education as it can undermine academic integrity. This study aims to analyze the distribution of Turnitin results in academic library documents from various study programs at Universitas Nasional, Jakarta. Using a data visualization approach and machine learning algorithms, this research explores the relationship between Turnitin scores and document validity status. The methods used include data visualization through Scatter Plots, Violin Plots, and Box Plots, with these visualizations utilizing Orange Data Mining as the data processing method. Additionally, a logistic regression algorithm is applied to classify documents based on Turnitin scores. Furthermore, the Chi-Square statistical test is implemented to evaluate the significance of the relationship between Turnitin results and document validity status. The findings of this study indicate that Turnitin scores exhibit significant distribution differences among study programs, with notable disparities between valid and invalid documents. Documents from certain study programs tend to have dominant scores in the 20-50% range, which serves as a critical threshold in determining document validity. This study provides in-depth insights into the patterns of academic document validity and offers a data-driven approach to improving the quality of academic evaluation in higher education. Additionally, this research is expected to serve as a foundation for academic policies in strengthening plagiarism detection systems, increasing transparency in evaluation, and promoting the development of more accurate and efficient document validation methods. However, this study has limitations regarding the scope of data used, particularly in terms of study program representation and external factors that may influence Turnitin scores. Further research is needed to examine pedagogical aspects and academic policies that contribute to variations in Turnitin scores across different study programs.

Review

This study presents a timely and relevant investigation into the critical issue of academic plagiarism, leveraging Turnitin data from Universitas Nasional, Jakarta. The authors propose a comprehensive methodological approach, combining robust data visualization techniques—specifically Scatter, Violin, and Box Plots processed through Orange Data Mining—with a machine learning algorithm, Logistic Regression, for document classification. The integration of the Chi-Square statistical test further strengthens the analysis by evaluating the significance of the relationship between Turnitin scores and document validity. This multi-faceted approach offers a promising framework for understanding plagiarism patterns and enhancing academic integrity, making a valuable contribution to the ongoing efforts to maintain high standards in higher education. The findings reveal significant insights into the distribution of Turnitin scores, highlighting notable differences across various study programs and distinct disparities between documents categorized as valid and invalid. Crucially, the research identifies a critical threshold, with many documents from certain programs exhibiting dominant scores in the 20-50% range, which is posited as a key determinant of document validity. These results provide an in-depth, data-driven understanding of academic document validity patterns. The study's implications are far-reaching, offering a foundation for improving academic evaluation quality, strengthening plagiarism detection systems, promoting transparency, and developing more efficient validation methods, thus potentially informing crucial academic policies. While the study provides valuable contributions, the authors rightly acknowledge certain limitations. The scope of the data, particularly concerning the representation of study programs and potential external factors influencing Turnitin scores, warrants further consideration. These limitations suggest that the generalizability of the findings might be constrained to the specific context of Universitas Nasional. The call for future research to delve into pedagogical aspects and academic policies that contribute to variations in Turnitin scores across programs is a pertinent suggestion, indicating a clear path for expanding this important work and offering a more holistic understanding of plagiarism in academic settings.

Full Text

Comments

You need to be logged in to post a comment.