SENTIMENT ANALYSIS OF ONLINE REVIEWS USING BERT

Computer Engineering

The use of the BERT model of neural networks to evaluate emotions from online consumer feedback has been effective in improving understanding of customer preferences for online products. This technique not only helps platforms better meet the needs of its customers, but it also helps customers find appropriate and affordable products. This development helps to improve the accuracy of product suggestions through intelligent processing. A number of sentiment analysis tests were carried out using the pre-trained BERT model. A highly accurate categorization model was created by making meticulous parameter tweaks throughout the experimentation phase. The BERT layer was used as the fundamental word vector layer. This layer received input text sequences and transformed them into vectors. After being processed by respective neural networks, the vectors were subjected to classification using the SoftMax activation function. Sentiment analysis entails evaluating customer attitudes using text data, notably product reviews in this instance. Our research led to the development of a solution proficient in discerning sentiments conveyed in online product reviews. To accomplish this, we utilized algorithms like Logistic Regression, Random Forest Classifier, and Sentiment Intensity Analyzer. The experimental results demonstrate the accuracy attained in conducting sentiment analysis.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.