An innovative word embedded and optimization based hybrid artificial intelligence approach for aspect-based sentiment analysis of app and cellphone reviews

A one-grained problem in Natural Language Processing (NLP), "Aspect-Based Sentiment Analysis (ABSA)" seeks to predict the sentiment polarity of several features in a sentence or document. The majority of the present research concentrates on the relationship between a given context and an aspect sentiment score. Inadequate attention has been paid to the significant deep relationships between the global context and aspect sentiment polarity. In this article, a novel word-embedded and optimization-based hybrid artificial intelligence (AI) method is proposed for ABSA of different customer review datasets. The review dataset was gathered in the initial stage using a web scraping algorithm. Here, the analysis is validated with the help of Flipkart Cell Phone Reviews and the ABSA Warehouse of App Reviews (AWARE) dataset. Using a pre-processing strategy, the raw data is improved as informative data. Additionally, the Convolutional Neural Attentive Bag-of-Entities (CNABE) of pre-trained word embedding is proposed, which provides the most efficient feature engineering and effectively preprocesses the words/characters for enhanced representation. Then, the Remora Optimization Based Extreme Action Selection Gradient Boosting (RO-EASGB) algorithm is proposed for sentiment analysis classification using the benchmark datasets. The implementation of this research is done using Python software. The performance of the proposed method is compared with the existing methods in terms of accuracy, recall, precision, F1-measure, and so on. Based on the experimental outcomes, the research shows that the proposed approaches outperform the existing state of the art methods.