PARAPHRASE EXTRACTION BY USING MACHINE LEARNING ALGORITHMS- A REVIEW

Abstract

Data mining, copyright detection, Authorship authentication, information extraction and text summarizing are among applications that require paraphrase detection. To recognize paraphrases that compare various corpus-based, and particularly deep learning (DL) models. Students' comments on subject forums are compared to the performance of standard machine learning algorithms and two deep learning techniques in obtaining subject keywords. Student comment data over the past two years was collected for this purpose, with a portion of the raw data being manually labelled. The present review focused on comparisons that included naive Bayes, logistic regression, support vector machines (SVM), artificial neural network (ANN), and Long Short-Term Memory with Attention (LSTM) (Att-LSTM). The four evaluation measures were used to measure the performances. This strategy is the most effective for extracting concept keywords from student comments, based on our experiment and visualization statistics. A certain degree of symmetry is also found in the findings of the algorithms and the depiction of the results. There is practically a reflection symmetry in the retrieved subjects from comments posted at the same phases of different training sessions.