Singapore University of Technology and Design, 8 Somapah Rd Singapore 487
When visiting unfamiliar cities, it is crucial for tourists to have a well-planned itinerary and relevant recommendations. However, many tour recommendation algorithms only take into account a limited number of factors, such as popular Points of Interest(Pois) and routing constraints. Consequently, the solutions they provide may not always align with individual user of the system. In this paper, we propose 2 iterative algorithms: PPoiBert and BtRec, that extend from the PoiBert algorithm to recommend personalized itineraries on Pois using the Bert framework. Firstly, we propose PPoiBert as a basic framework for recommending personalized itineraries, depending on different user inputs; secondly, our BtRec algorithm additionally incorporates users’ demographic information into the Bert language model to recommend a personalized Poi itinerary prediction given {𝑝u, 𝑝v}. Our recommendation system can create a travel itinerary that maximizes Pois visited, while also taking into account user preferences for categories of Pois and time availability. This is achieved by analyzing the travel histories of similar users. Our recommendation algorithms are motivated by the sentence completion problem in natural language processing (Nlp). We enhance the itinerary prediction using our proposed algorithm, BtRec (Bert-based Trajectory Recommendation,) that makes recommendations using trajectories with their demographic information such as cities and countries in the recommendation algorithm. The prediction algorithm in BtRec identifies a suitable profile that most similar to the query specification, before generating a list of Pois for recommendation. Using a Flickr data set of nine cities of different sizes, our experimental results demonstrate that our proposed algorithms are stable and outperform many other sequence prediction algorithms, measured by recall, precision, and ℱ1-scores.
