Abstract:
"Customer reviews are absolutely vital in today's digital world for determining a product or service's reputation. Businesses can obtain insightful feedback from their customers through a well-thought-out customer review system, which can then be used to improve products, services and to make more informed decisions. Customer reviews enables product manufacturers to improve their offerings and strengthen their relationships with customers, as well as the product consumers to make informed decisions.
Aspect-based sentiment analysis is a crucial task in the Natural language processing domain and in terms of customer review analysis, where the objective is to determine the product feature or aspect specified in a customer review and the sentiment indicated towards that aspect. In this paper, the author proposes a model for evaluating spans of aspects from customer reviews that determine the polarity, which has the potential to discover more about consumer preferences and perceptions. Proposed model uses a combination of natural language processing techniques and deep learning models to accurately identify and categorize aspects, sentiment, and opinion expressed in customer reviews where customers can obtain polarity scores for goods and services based on various aspects.
Author tries to discover a novel general approach to work through problems mentioned in related work, proposes a framework for aspect based sentiment analysis (ABSA) with the help of span based representations. A model with a new layering system that has not been done before to address the problem of contradictory sentiment polarities in real world scenarios all the while considering the local context in those mentioned scenarios which in turn, may also be the solution for filtering out wrong pairs of aspects and sentiments that contributes to displaying wrong predictions to help businesses make accurate decisions and for customers to get an idea of a particular product or a service. "
