Pengembangan Sistem Rekomendasi Rencana Perjalanan yang Dipersonalisasi Berbasis Text Mining dan Optimasi Rute
Salsabila Miftah R, Ir.Budhi Sholeh Wibowo, S.T., M.T., PDEng., IPM., ASEAN.Eng.
2025 | Tesis | S2 Teknik Industri
Optimal travel itinerary planning is a challenge for travel recommendation system developers. The main difficulty lies in extracting tourist preferences from unstructured review data and optimizing travel routes to align with destination operational time constraints. Most existing recommendation systems consider time only as an operational constraint, such as opening or closing hours, without incorporating tourists' personal preferences. Additionally, available route optimization algorithms, such as the Vehicle Routing Problem with Time Windows (VRPTW), have not explicitly integrated individual preferences, resulting in generic recommendations. This research aims to bridge this gap by developing a travel recommendation system that incorporates tourist preferences into route optimization.
This study aims to develop a personalized travel itinerary recommendation system by combining text mining techniques to extract tourist preferences from online reviews and the VRPTW algorithm to optimize travel routes. The primary research question addressed in this study is how to integrate tourist preferences into route optimization to generate more relevant and efficient itineraries. This study employs a data-driven approach with two main stages: text mining to identify tourist preferences from online reviews using rule-based classification and travel route optimization using a heuristic VRPTW algorithm implemented with Google OR-Tools.
The findings indicate that the proposed model can extract tourist preference information from online review data and classify it based on the time-of-visit dimension. As a result, the system can provide travel destination recommendations that align with users' preferred visit times. Additionally, the system successfully enhances travel efficiency by minimizing total travel distance while ensuring that recommendations remain realistic within the operational time constraints of destinations. Model validation using the Solomon dataset shows an average travel distance deviation of 30.54% compared to the benchmark solution, with the smallest deviation of 11.13% in simple scenarios and the highest deviation of 54.86% in more complex scenarios with strict time constraints.
Kata Kunci : Text mining, optimasi rute, VRPTW, rekomendasi perjalanan wisata, e-tourism, perencanaan perjalanan