Abstract:
Today's world is easily depended on technologies. Nowhere is found running fast & without
digital compartments. Even peoples’ directly approaches are covered through the internet
which made our lives not only easier but also accruable. Online shopping platform is growing
faster among the whole world as well as our digitally developing country. Many e-commerce
shopping sites are taking place of our day to day marketplace as we can purchase any
available material or service staying at home.
Despite of the benefits there are also critics of e-commerce since the customers leave a
review about the service of products in text review form. Sentiment analysis of online
reviews has been researched ago but on Bengali language there are not found very promising
work. So we found a scope of research on this topic to classify the sentiment behind
Bangladeshi e commerce sites’ reviews and deliver a modeled output. This will help the new
customers to acknowledge a true description of old buyers and also the sites can upgrade their
service or product through customer satisfaction. We used three different algorithms to sort
our data to see which one gives better accuracy to human sentiments and classify the reviews
as excellent, better, good, poor or very bad.