KomoTrip: a multi-day travel itinerary recommendation method based on the discrete komodo mlipir algorithm

Area of Science:

• Artificial Intelligence
• Operations Research
• Computer Science

Background:

• Sophisticated recommender systems are crucial for the Tourist Trip Design Problem (TTDP), often modeled using analogies like the Team Orienteering Problem with Time Windows (TOPTW).
• Existing TOPTW approaches for TTDP lack personalization through user preference weighting and face challenges with computational efficiency.
• The Komodo Mlipir Algorithm (KMA) offers promising scalability for optimization problems.

Purpose of the Study:

• To propose KomoTrip, a novel method for personalized multi-day travel itinerary recommendations.
• To address the limitations of existing TOPTW models in TTDP by incorporating user-weighted multi-attribute preferences.
• To improve the computational time efficiency of generating optimal travel routes.

Main Methods:

• Developed KomoTrip, integrating a discrete version of the Komodo Mlipir Algorithm (KMA) with Multi-Attribute Utility Theory (MAUT).
• Implemented a multi-attribute approach to accommodate user preferences for personalized route optimization.
• Conducted evaluations using general performance scenarios, Degree of Interest (DOI) combinations, and varying Points of Interest (POI) counts.

Main Results:

• KomoTrip demonstrated superior computational time efficiency compared to benchmark algorithms across various problem scales.
• The method achieved robust fitness values and competitive profit values, especially for longer tour durations.
• Consistent superior runtime performance was observed when benchmarked against state-of-the-art TOPTW heuristics on the Solomon dataset.

Conclusions:

• KomoTrip is an efficient algorithm for recommending optimal multi-day tour routes.
• The method effectively incorporates weighted multi-attribute user preferences into the optimization process.
• KomoTrip offers a scalable and time-efficient solution for personalized tourist trip design.