Perpustakaan Digital ITB

Instagram has now become one of the main channels for promoting tourist destinations, driven by the active role of influencers in shaping public perception. This study designs an Instagram bot named Nusava, a prototype social media bot developed to leverage the behavioral patterns and popularity of tourism influencer accounts. Nusava consists of four main modules: the data mining module, the Large Language Model (LLM) module, the dashboard module, and the automation system module. The data mining module is responsible for providing general information about tourist destinations, tourism destination patterns, and influencer behavior. Destination information is obtained through scraping from the Tiket.com and Traveloka platforms. Tourism destination patterns are analyzed using similarity measurement with the cosine distance function. In addition, the data mining module also calculates the geographical distance between tourist attractions, so that destinations that are geographically close and share similar characteristics can be recommended. To obtain influencer behavioral pattern data, user interaction data on Instagram needs to be stored in a social network, which is then divided into several communities using the Speaker-Listener Label Propagation (SLPA) algorithm. Within each community, the users who are influencers in that community and the popular posts will be analyzed. Afterwards, the data mining module will be integrated with the LLM module to provide labels and brief descriptions of these communities. This study mined 1,348 tourist attractions, formed a social network containing 479,442 users and 2,287 posts, and discovered 290 communities. Through testing and evaluation, the results show that an Instagram social media bot based on LLM and data mining can support the promotion process of tourist attractions.