Aspect-based sentiment analysis of mobile phone reviews using LSTM and fuzzy logic

Nowadays, reviews about the products on online shopping sites become an essential source to help the customer to take better decisions on buying a product and achieve good sales of the product. It becomes a habit for the consumers to share opinions about a recently purchased product immediately on online shopping sites and social media websites. So, there is a huge demand for an intelligent system to detect sentiments from customer reviews under specific aspects of a product posted on online shopping sites. In recent years, various machine learning techniques have been experimented on various benchmark datasets to analyze sentiments expressed by the consumer through online portals. But, consumers are still struggling to get aspect-based sentiments expressed by other consumers, and the accuracy of the existing model is not satisfactory. Hence, we proposed an intelligent system using long-term short memory with fuzzy logic to classify consumer review sentences under various aspects with four different labels, namely highly negative, negative, positive and highly positive. So, consumers who wish to buy a new product from online portal can see multi-label sentiments of the various aspects of the product quickly. The proposed system was experimented on Amazon cell phone review, Amazon video games review and consumer reviews of amazon products benchmark datasets and obtained the results with the accuracy of 96.93%, 83.82% and 90.92%, respectively. The proposed model outperforms in terms of accuracy when compared to the state-of-the-art-methods. The developed system also analyzed the product reviews based on the current trends and geographical location. The proposed system aids the manufacturers for improving the products based on customer complaints.