Recommendation System for Sightseeing Tours

Recommendation System for Sightseeing Tours

Ricardo Claudino Valadas (Instituto Superior Técnico de Lisboa, Lisboa, Portugal) and Elizabeth Simão Carvalho (CIAC/UAb, University Aberta, Portugal)
DOI: 10.4018/IJTHMDA.2020070104
Article PDF Download
Open access articles are freely available for download

Abstract

This research proposes a model of a recommendation system (RS) for tourist itineraries. The RS suggests tips of what to visit in a city, based on the available time, personal preferences, current geo-location, and the user's context awareness. These suggestions are calculated based on the treatment of collected data in real time by external application programming interfaces, through a list of points of interest located within a radius that can be reached by the user. Preliminary tests validated the model's goals and its potential in the tourism sector. The RS for tourist itineraries proposed is based on four essential points, in order to make the experience different and well as possible: end-user's personal tastes, the time available, end-user's current location, and context awareness. The performance tests that were carried out brought very positive results and showed that the RS presented a number of requisitions proportional to the server response times and algorithm. The functionality tests were quite positive, with percentages of experience of using the RS between 62.5% and 100%.
Article Preview
Top

Background

PAT-Planner is a project that aims to personalize a route based on points of interest (POI) nearby, available time and budget of the user (Lu et al. 2016). Its concept aims to address the planning of a trip combining existing tourist packages and created by travel agencies, with tourist attractions included. It demonstrates the creation of a new RS that allows to generate a tourist itinerary in a zone based on the user's personal interest, available time and maximum budget. Several users’ tests have been conducted and presented proving the consistency and functioning of the PAT-Planner.

Multi Request Route Planning (MRRP) is based on the Dijkstra algorithm (Dijkstra 1959). The authors of this article want to suggest to the users a walking route through a different types of places in urbans environments (Lu et al. 2016). CA is not a concept mentioned by the authors, but this RS ends up using its essence (figure I). The problem is that is not taken advantage of, for example: The user knows which direction should take and which stores can go to, but at no point is it mentioned in the study that the RS takes into account stores that are currently open or closed.

Figure 1.

Multi request route planning (Source: Lu et al. 2016)

IJTHMDA.2020070104.f01

Trip planning route main goal is to suggest tourist routes according to the working hours and the popular time at each place to visit (Chia et al. 2016). According to the coordinates of a certain point to visit (latitude and longitude), hours of operation and recommended time of these same sites, this RS for tourist itineraries suggests to the user what to visit on the island of Penang, Malaysia.

Xenia is the name of the recommendation system for tourist itineraries created by Korakakis, Mylonas & Spyrou (2016). It aims to receive the data that come from the social network Flickr, which inform the system about POI in the vicinity, also taking into account the context-of-awareness (CA) concept. Information on the activity of the user in this social network is considered as the point of the origin of the information.

Trip-Mine wants (Lu et al. 2011) to improve the recommendations in the tourism sector according to the user's location. The biggest challenge is to deal with the user's available time. The authors studied and departed from the Travelling Salesman Problem (TSP) algorithm. This is a positive point because this algorithm is fully suited to performance problems in the search for the quickest way to go.

Complete Article List

Search this Journal:
Reset
Volume 6: 1 Issue (2022): Forthcoming, Available for Pre-Order
Volume 5: 2 Issues (2021)
Volume 4: 2 Issues (2020)
Volume 3: 2 Issues (2019)
Volume 2: 2 Issues (2018)
Volume 1: 2 Issues (2017)
View Complete Journal Contents Listing