Abstract:
Sentiment polarity detection has recently piqued the interest of NLP researchers, owing to
the resulting in positive of consumer comments or ratings on the internet. Due to the
continued expansion of e-commerce sites, the rate of purchase of various products has
grown. For example, people's interest in books is fast growing. Online marketing and ecommerce companies were already prospering in Bangladesh during this era of internet
technology. For example, product reviews on the Internet have become an essential source
of information for customers making purchasing decisions. Because there are sometimes
too many reviews for consumers to read, figuring out how to automatically classify and
determine sentiment from them has become a major research challenge.
Books are said to be a person's best friend. Books are crucial in every human's life because
they provide information of the outside world, improve their reading, writing, and speaking
abilities, and improve memory and intellect. Even just a few years ago, individuals in
Bangladesh had to travel to the library in person to get books. Many of the benefits are
straightforward, and the internet bookshop has a disadvantage in that the reader is
unfamiliar with the books or with the book store itself. To avoid this, book readers prefer
to rely on reviews and ratings. Our goal is to assess Bangladeshi language reviews and give
accurate information about books and online bookstores so that book lovers may buy the
right books to read and find better online bookstores. In this paper we show how to extract
sentiment polarity (positive or negative) from Bengali book reviews using machine
learning and Natural Language (NLP).We used five classification methods : Adaboost
,Decision Tree (DT), Random Forest Tree (RFT), Support Vector Machine (SVM), and
lightGbm algorithm.
