ROBERTA TIL MODELI VA NMF TEMATIK MODELLASHTIRISH USULI

Abstract

Online reviews now provide customers with relevant information for the purchase decision process. Such reviews change the way online consumers make decisions and form their impressions of a product, as well as affect the effectiveness of product sales. According to Consumer Reports, a product without online reviews is purchased 270% less often than a product with 5 positive online reviews. In addition, a public survey shows that 98% of online consumers use online reviews as information when making online purchases and spend more than a minute reading online reviews before making a purchase. However, buyers are often skeptical of reviews that are too numerous. Studies show that 82% of customers specifically look for negative reviews. It can be concluded that reviews, in turn, play a very important role in making a purchase decision. Therefore, the following is an analysis of reviews, their sentiment analysis using various methods, to determine what prompted the customer to make such a review, and to provide product providers with the right solution for each case. Recent studies have worked in the areas of sentiment analysis, comparing the different available methods, and identifying the methods that give the best results, but none of them have made significant progress in identifying the subject matter of reviews left by customers. This is where the uniqueness of this work comes in. This study not only determines whether a review is positive or negative, but also models the subject matter of the reviews. We find answers to the questions of identifying the elements that affect customer satisfaction and how to make the right decision in each case.