Sentiment Analysis and Topic Modelling of Tourist Reviews on Heritage Destinations of Lawang Sewu

Abstract

Unstructured online reviews for heritage destinations like Lawang Sewu offer a rich source of data for an in depth analysis of visitor perceptions. A combined approach of supervised sentiment classification and topic modeling was applied to a bilingual review corpus to identify the specific drivers of satisfaction and dissatisfaction. The analysis revealed a clear dichotomy in visitor feedback. The overall sentiment is overwhelmingly positive. This positivity is driven by the site's core heritage value, its historical architecture, and a family friendly atmosphere. In contrast, negative sentiment is highly concentrated and not directed at the core attraction, but rather at specific operational frictions. These frictions are primarily related to ticketing, pricing, and the perceived quality of guided tour services. This integrated approach provides management with actionable, data driven insights, enabling them to protect celebrated strengths while systematically addressing logistical weaknesses. This work demonstrates how unstructured public feedback can be transformed into a strategic tool for targeted service improvement at heritage destinations.