High-performance sentiment classification of product reviews using GPU(parallel)-optimized ensembled methods

Abstrak

Sentiment analysis is an important approach in natural language processing (NLP) that extracts information from text to infer underlying emotions or views. This technique entails classifying textual information into feelings like "positive," "negative," or "neutral." By evaluating data and labeling, client input may be classified on scales such as "good," "better," "best," or "bad," "worse," resulting in a sentiment classification. With the fast expansion of the World Wide Web, a massive library of user-generated data—opinions, thoughts, and reviews—has evolved, notably for diverse items. E-commerce firms use this data to gather attitudes and views from social media sites like Facebook, Twitter, Amazon, and Flipkart. The GPU-CUDA-ENSEMBLED algorithm is a GPU-accelerated method for sentiment classification, enhancing predictive performance by minimizing variances and biases. It outperforms existing algorithms like SLIQ and MMDBM, demonstrating GPU mining's efficiency. The proposed algorithm utilizes GPU-accelerated sentiment analysis to accurately predict smartphone ratings, providing valuable insights for businesses to maximize customer feedback potential.