DATA-DRIVEN INSIGHTS ON POST-COVID MSME RECOVERY: INTEGRATING TEXT AND REVENUE ANALYSIS IN WEST SUMATRA

Abstract

This research examines how media narratives frame the condition of MSMEs during the crisis and post-pandemic recovery by utilizing topic and sentiment analysis from 1422 online news stories containing the keyword MSMEs. The main problem raised is the lack of empirical understanding of how media reports reflect changes in MSME performance and whether the dynamics of these narratives move in the same direction as business recovery. This research aims to map topic patterns, measure news sentiment using a special MSME–crisis sentiment dictionary, and link it with MSME income trends as an business indicator. Through the stages of text preprocessing, LDA topic modeling, and lexicon-based sentiment calculations, this research found that in the early phase of the pandemic, the news was dominated by crisis topics and had a negative tone in line with the decline in MSME income. Entering the recovery phase, the proportion of government support topics, economic programs and market activity increased, followed by a shift in sentiment to become more positive. In the post-pandemic period, the media narrative has become increasingly stable and growth-oriented, consistent with the trend of increasing MSME income. These findings show that the dynamics of media narratives move parallel to the economic conditions of MSMEs and can be a supporting indicator for reading the crisis, transition and recovery phases. In conclusion, text mining in local news can provide strategic insight for local governments and MSME actors in understanding public perceptions and directing communication strategies and economic recovery policies.