<![CDATA[In this digital era, Instagram is not only the most popular photo and story sharing platform, but also acts as a source of important and critical information such as information about emergency situation. This research aims to utilize caption data from Instagram posts to produce a map showing the location of emergency situations in South Jakarta. From 1088 captions, 204 of them were successfully identified as informative and relevant texts for this study. The informative data then processed and visualized as a geographic information system using the OpenStreetMap map and the Overpass Turbo API, so that the resulting information can be more easily understood and can be used as material for further analysis by the public or authorities. This research combines the TF-IDF and Multinomial Naïve Bayes methods for classifying informative or uninformative captions on Instagram posts, as well as the Levenshtein Distance Algorithm for cleaning the dataset by correcting typos to reduce dimensions and improve the quality of the data that are going to be analyzed. The challenge in overcoming data imbalance is by implementing three text augmentation techniques, namely synonym replacement, random swap, and random deletion. From the model training process, an accuracy rate of 92,6% was obtained. This result shows that the proposed method is not only effective in improving classification accuracy, but also successfully provides informative visualizations related to the location and frequency of emergencies in South Jakarta.]]>
Copyrights © 2025
