Consumer Sentiment Analysis to E-Commerce in the Covid-19 Pandemic Era

Abstract

This study aims to identify e-commerce problems during the Covid-19 pandemic as a basis for providing recommendations for improving e-commerce services. The methods used in this research are Naïve Bayes Classifier (NBC), Text Association, and Focus Group Discussion (FGD). The NBC method is used to classify consumer sentiment, while the Text Association is to find the relationship between words. The data source used is consumer reviews submitted on Twitter for the period January-April 2021. The FGD method is used to classify the results of the Text Association into service marketing mix elements, identify root causes using Fishbone Diagrams and develop recommendations for improvement. The respondents involved in the FGD were e-commerce experts representing practitioners and academics. The result of the improvement recommendation is: educating sellers always to provide accurate product specification information and implementing quality control; increasing mutually beneficial cooperation with logistics service providers, especially concerning service commitments and cutting shipping costs; improve information disclosure of promotional events; improve the refund procedure; optimization of multichannel e-commerce software; increase accuracy in the process of selecting and verifying prospective seller data; improve transaction process monitoring; optimizing the application of a reward and punishment system based on consumer reviews.