SENTIMENT ANALYSIS BASED A NOVEL HYBRID RECOMMENDATION SYSTEM

Abstract

Today's technological advances and the globally effective Covid-19 pandemic have increased the importance of e-commerce sites and stores have offered their products to users through e-commerce sites. Especially in march, april and may, when the pandemic started to be effective in our country, according to the Ministry of Commerce data, e-commerce sites grew by 19% compared to the previous year, and the number of orders they received increased by 292 million during the pandemic period. However, with the popularization and widespread use of e- commerce sites, it has become more difficult for users to make the most appropriate choice among a large amount of products. For this reason, systems that enable people to obtain products from specialized information have gained importance. In this way, it allows the user to both deal with less data and reach the product they are looking for in a shorter time. This situation has increased the importance of recommendation systems and made them one of the popular topics today. Recommendation systems can be defined as systems that aim to recommend new products that users have previously interacted with other users similar to them have interacted with as reference. In e-commerce sites, the comments made by other users on the products are one of the most important factors that users refer to when choosing products. In this study, a new hybrid recommendation system has been developed that classifies and analyzes the comments made on products as positive, negative or neutral with sentiment analysis methods and recommends them to other users accordingly.