Analysis of Syntactic and Word Vector Using Cross Language Translation Algorithm

Abstract

Digital technologies have created the need to preserve intellectual originality in translated works has arisen as a result of digital technologies. To measure the degree of resemblance between textual pairs created in various languages and identify plagiarism, a cross-lingual syntactic link is essential. This study investigates plagiarism detection and evaluation, as well as cross-lingual syntactic analysis, using a dataset of university students. The Natural Language Process (NLP) imports the data into a research model for translation and plagiarism detection. We then extract the data into features using a word vector. The study suggested using the term embedded fast recurrent network (WE-FRN) for plagiarism detection and syntactic analysis. The study presented numerous neural network designs to address the problem of plagiarism detection, including binary classification (syntactic regression analysis of documents) and regime proposals (plagiarism or independently authored). The experimental results showed that using WE-FRN with rich syntactic characteristics yielded better results than the baseline. Additionally, we used the regression of the source and suspicious documents to analyze the classification's loss function. We trained the dataset using a five-fold accuracy of WE-FRN, yielding the highest accuracy values for English (90.9), Chinese (88.87), Japanese (87.49), and Korean (87.26).