Exploring Indonesian e-wallet usage and category classification in social media using geocoding through X analysis

This study visualizes Indonesian e-wallet tweets combining Social Network Analysis (SNA), geocoding and classifying data tweet to deepen marketing insights. SNA serves as a business intelligence tool, while geocoding with named entity recognition (NER) and ArcGIS are used to predict the point of position in data spatial. StanfordNER tools are used to build a location classification model using NER with six classes, include province, district/city, sub-district, village, road, and place names. Each tweet was also classified using two aspects using IndoBert, sentiment (positive, negative, and neutral) and categorization (features, access, service, security, and responsiveness). Thus, each tweet underwent several tasks: sentiment classification, category classification, and finally, spatial data points. The tweet data was then created into an integrated map to facilitate analysis of e-wallet usage in Indonesia. This research uses social media “X” as known “Twitter”. Neutral sentiments dominate, showing that conversations are mostly informative. E-wallet features and services are the most discussed aspects, especially in capital city, Jakarta and crowded island (Java), emphasizing the need for improvements. Negative sentiment is concentrated in West Java and Jakarta, mainly due to security and service concerns. Seasonal trends highlight engagement spikes during promotions and decline due to unmet expectations.