Sentiment Analysis of Social Media Discourse on Electric Vehicles in The Kyrgyz Republic

Abstract

Air pollution is a prominent issue causing numerous harmful effects on health, ecology, and climate change. Transportation is considered the second-largest contributor to air pollution. The current focus on electric vehicles is the promising decision in reducing carbon dioxide emissions. The Kyrgyz Republic indicates strong intention to foster promotion of electric vehicles. Although the number of electric vehicles increases, statistics demonstrate insignificant share of electric vehicles in the country. Hence, this thesis is aimed to analyze public perception of citizens using an online social network (OSN). Instagram is the most popular OSN in the Kyrgyz Republic, thus, it was decided that it would be the source of data. Public perception is captured utilizing two widely used models for sentiment analysis, VADER and BERT. The thesis emphasizes the high accuracy of BERT model and efficiency of pseudo- labelling method in training a model. The results of the models indicate a dominance of neutral sentiment. Moreover, it is noted that negative sentiment slightly outweighs positive sentiment. News related to infrastructural construction for electric vehicles were almost always met with positive sentiment. This could indicate citizens’ intentions to switch to electric vehicles if infrastructural conditions are satisfactory. Thesis also analyzed relationship of positive sentiment and the number of registered electric vehicles. It showed the sufficiency of positive sentiment at a 20% base in order to keep the gradual adoption of electric vehicles. Moreover, study indicated that public perception might not be the major factor of adoption of electric vehicles. In addition, this thesis conducted differential lexical analysis of positive and negative comments to derive specific factors causing particular sentiment. Finally, the thesis analyzed content of the comments by manual reading and presented summary of popular opinions on electric vehicles.