Generative Paraphrasing And Predictive Sentimental Analyser Using Novel Restbert Framework

Abstract

Sentimental analysis mainly categorizes texts according to their polarities. One of the critical areas of text mining is sentiment analysis. The various opinions of various users within every given category are compiled into a single dataset and considered for examination. Generally, any machine learning technique could be used to aid in the study. It aids in the text's efficient categorization. This study uses various word vectorization approaches to identify paraphrasing in phrases. Text-based data can be extracted from a big data collection via vectorization. This is done by linking the word with a vector. Due to the size of the information in text format, it needs to be defined This problem can be solved using simple to complex techniques. Using BERT, we compare word vectorization approaches in this work. We have used three different datasets for the comparison. By categorising the performance characteristics into five separate groups, we were able to achieve an accuracy of 95% for the metrics of accuracy, recall, precision, and f1.