Analysis of Content Consistency in Scientific Journal Based on Natural Language Processing and Machine Learning

Abstract

This groundbreaking research proposes implementing a state-of-the-art monitoring system designed to evaluate the structural cohesion of scientific journal manuscripts and generate accurate sentence interpretations. By utilizing Natural Language Processing (NLP) and the power of Machine Learning (ML), this research aims to answer how authors can know their consistency in writing papers by self-detecting. This system provides accurate comparative analysis results and sentence interpretation ratios to measure author consistency in writing journal manuscripts or papers. By using natural language processing (NLP) as a preprocessing stage and applying the Term Frequency-Inverse Document Frequency (TF-IDF) as a determinant for vectorization by dividing two vectors, this study uses Support Vector Machine (SVM) for predictive classification in machine learning. It uses Cosine Similarity (CS) to distinguish the similarity of sentences. The results were staggering: the study achieved an 83.94% accuracy rate for relevance consistency in content comparison analysis, supported by the activation of 2485 journal datasets, with a yield of 0.740402 obtained from convolution optimization. This remarkable achievement has the potential to revolutionize academia by improving the efficiency and quality of writing, providing easy-to-understand information for novice researchers trying to write scientific papers. By implementing this monitoring system, researchers can ensure that their manuscripts are structurally cohesive and consistent and can enjoy the benefits of a more efficient and smooth writing process, resulting in better quality research and more impactful publications.