| dc.description.abstract | E-commerce platforms have increased user satisfaction, they have also led to the rise of customer apprehension during the online purchases. Physical stores offer one important benefit to customers that e-commerce lack, which is the ability to experience the object before making a purchase decision. For a lot of the daily and essential goods, this does not cause any problems as people are well aware of what to expect from the goods, some example could be toothpaste, bottled water, packaged snacks, etc. However, there are some goods whose value can only be fully determined after experiencing the objects. This could include books, electronics, furniture, etc. The inability to experience these goods led to lower online purchases due to the added concern of realizing poorer quality after a purchase is made. E-commerce platforms have tried to alleviate this concern by adding images, descriptions, and reviews by other customers. The reviews have become a means for past customers to state their opinions as well as an important source of information for new customers. Being created by customers, these reviews can grow very quickly in short periods of time, and it has become a challenge for the platforms to figure which reviews to show to the customers. Many players have tried different ways to rank the reviews and place the most effective ones at the top to encourage purchases, or to provide the most information. However, given the unstructured nature of this problems, robust solutions are yet to be developed.
Our aim in this study to understand how reviews can be analysed for informational content. Our focus will be to understand what characteristics of a review make it more likely to be helpful for the customers during their purchase decision process. While ML/AI models would work best to analyse and understand the contextual nature of reviews, we believe that important insights can be drawn with simpler and more intuitively developed characteristics as well. | en_US |
