Consumer Reviews in the E-commerce Market: Features of Semantic Analysis

The e-сommerce industry is growing rapidly, and online stores are becoming more involved in people's daily lives. Online reviews, according to previous research, strongly affect product sales. Therefore, online reviews have become a сentral research topic for many marketing studies. The fundamental part of this research is the analysis of the online reviews from apparel sector, using semantic content analysis. The research question of this study is: «How are the semantic and essential characteristics of online product reviews interconnected? » The aim of this thesis is to observe, what content do reviews include, what characteristics do they have, and how they may change depending on various factors. With the implementation of semantic analysis, semantic characteristics were highlighted and later compared with essential characteristics in order to find how are they interconnected. In frames on this research, nonparametric tests were conducted to either reject or confirm hypotheses, stated in the research. For the most of it, these hypotheses aim to establish reviews’ characteristics and analyze the interrelations between them. Characteristics such as review length, helpfulness, star rating, and sentiment were checked for the existence of relationships, as well as for the distribution of subcategories of characteristics within other characteristics. Also, a topic modelling was implemented that highlights the main comment topics among the star categories, allowing you to see the difference in the topics raised by customers regarding different shopping experiences on the Amazon.com website in the apparel sector. This research paper’s design is helpful to increase understanding of customers’ preferences, company’s strong sides and flaws, as well as to be more accurate in business-decisions.