Cruise tourism can be easily characterized as an “experienced” generator industry noting that the services provided by the cruise liners have experiential core while their consumption aims at delivering certain emotional outcomes. The cruise experience is comprised of two components: the onboard experience and the onshore.
The objective of the “Destination Challenge” is to enhance cruise passengers’ onshore experience through providing a blending of cultural activities and local production consumption activities, linked through a treasure hunt game.
A simple map-based app will challenge cruise visitors to experience the destination while benefited by the special offers and discounts -called “crusemons”
Destination Challenge: What problem are we trying to solve?
Cruise tourism expenditures in destination ports are composed of a broad range of spending including:
- Onshore expenditures by passengers which are concentrated in shore excursions and retail purchases of clothing and jewelry and purchases of food and beverages;
- onshore expenditures by the crew which are concentrated in purchases of food and beverages in restaurants and retail purchases of jewelry and clothing;
- expenditures by cruise lines for port services, such as dockage fees and linesmen, utilities, such as water and power and navigation services; and
- purchases of supplies, such as food and other supplies, by the cruise lines from local businesses.
To that extend the main challenge addressed is to increase the amount cruise passengers spend in local level and develop experiences.
Our technology is based on Recommender Systems. They consist of a tool for data analysis in order to match items with users and their specific needs. This merging scientific field, counting less than twenty years, is constantly growing with numerous new techniques and applications. Destination challenge is taking the advantage of a novel graph-based algorithm for recommendations.
How we do it? Firstly, we examine some notable results from node ranking and random walks on graphs. Then we use the notion of functional rankings and we proceed with the Multi-Damping framework for ranking. We use a new framework that takes in consideration a lot of aspects for a more complete manipulation of the data. Our approach is graph based. Rankings are produced by random walks with restarting, where the number of steps is determined by spectral properties of the transition probability matrix. In each step we focus on the closest neighbors and a re-ranking step takes place in the end which offers diversity.