Opinion Mining of Heterogenous Social Media Reviews on the Proposed Adoption of eNaira

Abstract

The Nigerian eNaira version launched recently was aimed at deploying digital currency services for the Nigerian monetary market. The opinion of Nigerians towards the adoption of the eNaira was conducted in this study. The methodology entails eNaira review acquisition from opinion mining reviews across heterogeneous social media platforms, text pre-processing, and sentiment analysis of tokenized review corpus. A dataset of 29505 unigram textual corpus was analysed to determine the positive, negative, and neutral sentiments expressed in their polarities. The pre-processing involved stop-word removal, stemming, lemmatization, Part-of-Speech tagging, entity tagging and tokenization. The opinion mining methodology deployed the Valence Aware Dictionary and Sentiment Reasoner (VADER) approach for the sentiment analysis of the eNaira tweets. The experimental results showed 56.25 %, 22.76% and 20.98% neutral, positive, and negative sentiments, respectively towards the adoption of the eNaira. The unigram analysis of the tokenized corpus likewise returns tokens including ‘app’, ‘bvn’ and ‘password’ as the most prominent unigram tokens while the bigram analysis returns (‘reset’, ‘password’) and (‘sent’, ‘email’) as the most prominent 2-word phrase with highest information gain. The result shows that eNaira enjoys favourable consideration for electronic commerce transaction.