Tourism is a rich and growing activity in the world. The digital support of the tourist experience is typically based on Tourist Recommender Systems (TRS). Widely spread TRSs typically recommend the visit of a set of Point of Interests (POIs). Unfortunately, some of these have also proved inappropriate because they do not consider in detail the indoor visit of a POI, such as a museum, which may contain several artworks and may have limited or regulated access. In this paper, we present our ongoing work on a Context-Aware Touristic Recommender System (CATReS). It provides an outdoor recommendation to select POIs suitable for a visit, as well as an indoor recommendation to select the artwork to visit. The overall recommendations are based on contextual information coming from both visitors' profile and the POIs' contextual information (e.g., the waiting time to access the museum, the average time to visit it). Our tour recommendation system is available and is being experimented at the Ufizi Galleries (Florence, Italy).
• Software and its engineering → Contextual software domains; Software implementation planning; • Information systems → Social recommendation. Context-aware planners; tourist recommender systems; outdoor and indoor visit.
The World Wide Web provides tourists with huge possibilities to search for interesting information and planning their activities. However, the large amount of data available makes it dificult to decide. Based on that need, Tourist Recommender Systems (TRS) are ofered by many resources to support tourists in managing their long, medium and short tours. A TRS can select and filter the relevant results to the user from a large database of tourist services including transportation, accommodations, attractions, Point of Interests (POIs), and even fixed tour package.
Some approaches focus on the concept of context, largely
associated with the delivery of service based on geographical information
and weather conditions [
The goal of this paper is to introduce CATReS, a Context-Aware Tourist Recommender System for the indoor and outdoor museum tour. CATReS, by taking in input both tourists contextual information (their user profile and preferences, current location, time availability) as well as cultural POIs contextual information (waiting time to access the museum, presence of artwork, museum location), generates a tourist plan at both macro-level (recommending which museums, among many, to visit in relation to the tourist time availability and preference, as well as waiting time to access it) and micro-level (recommending which specific artwork to visit in a museum).
Diferently from other approaches, CATReS, when generating a touristic plan, takes into explicit account cultural POIs contextual information (together with tourists contextual information), and generates a recommendation of feasible museums to visit, as well as the indoor visit plan.
The remainder of this paper is organized as follows: in Section 2 we briefly discuss the state of the art in tourist recommendation systems. Section 3 provides our initial work towards the realization of the CATReS system. Section 4 reports on an application scenario. Finally, conclusions are provided in Section 5. 2
In the literature diferent touristic context-aware recommender
system was proposed. Most of them use Orienteering models [
In [
To match the characteristics of tourism and leisure resources or
attractions with the user needs, TRS uses diferent
recommendation techniques. In [
CATReS is a Context-Aware Tourist Recommender System for indoor and outdoor museums’ tour planning. It consists of three main components: i) the context model, ii) the outdoor recommender system, and the iii) indoor recommender system.
The context model supports the definition and specification of contextual information. The context model takes into explicit account both visitor’s/tourist’s contextual information, as well as cultural POIs contextual information.
The outdoor recommender system, by taking in input the context model, recommends the cultural POIs that match the visitors’ profile and time constraints.
The indoor recommender system suggests the artworks to visit (in the selected museums) to (again) match the visitors’ preferences and time constraints.
Figure 1 graphically summarizes the expected output: given a set of museums with contextual information, and a visitor with her user preferences (the context model), CATReS returns a list of museums to visit (the outdoor recommendation), as well as the artwork to visit (the indoor recommendation), so to satisfy the imposed constraints (time for the visit, entry time to museums, etc.).
The focus of this paper is on the first two components, that are currently under deployment. users (with similar profiles) spent watching the artworks, and the last concerns the waiting time estimated to enter into each museum. • Weather conditions: we consider this information to suggest indoor/outdoor sites according to weather conditions (e.g., avoiding recommending to visit a garden in case of adverse weather conditions). 3.2
The outdoor tour recommendation can be formulated as a Tourist
Trip Design Problem (TTDP) [
∀i, j = 1, . . . , N ; ∀p = 1, . . . , P ti j xi jp ) ≤ TMax ; ∀p = 1, . . . , P
Finally, we added new constraints to model the user’s manual selection of POIs, to exclude from the outdoor paths sites in case of rainy days, and to exclude to visit a POI several times when more time-slots are available. 4
The CATReS outdoor recommendations approach is currently under deployment at the Ufizi Galleries (Italy), a museum complex in Florence comprising the Ufizi Gallery, the Pitti Palace, and the Boboli Gardens.
Those three sites are under the same administrative management and can be easily reached on foot. The Ufizi Gallery is the main museum, attracting several thousands of visitors per day. The Pitti Palace includes four museums: the Treasury of the Grand Dukes, the Palatine Gallery with the Imperial and Royal Apartments, the Modern Art Gallery, and the Museum of Costume and Fashion. Directly behind Pitti Palace are the marvelous Boboli Gardens.
The ongoing deployment process started in August 2019, uses the CATReS outdoor tour recommendation with a restricted set of contextual information including the timeslots, POIs, and Weather data (the three most right contextual information in Figure 2). By knowing the available entry time to access Ufizi and Pitti, the distance between those sites, the total time available for the visits, as well as the minimum reasonable visit time. In the optimization model, we considered the following parameters associated to the three sites: a museum score calculated as the average of user ratings on Google and Trip Advisor, the estimated time for visiting each site, and the time windows (timeslot in which the site can be visited).
Figure 1 shows an example of a resulting path. We assumed that the tourist buys the ticket at the Ufizi Gallery and that the first entry timeslot available is at 3:15 pm. The resulting path calculated by the optimization model implemented, considering the number of available tickets for the other museums is showed in figure. 5
In this paper we presented CATReS, a Context-Aware Tourist Recommender System that based on context, tourist preferences, and profiles, suggests indoor and outdoor tours. So far, we implemented a part of our model, that is being experimented in the Ufizi Galleries. We implemented a Team Orienteering Problem with a Time Windows optimization model that takes into account environmental context referred to weather conditions and site congestion. Future work will consider tourist preferences and profiles.
The authors would like to thank the Ufizi Gallery that is supporting this work under the agreement with the University of L’Aquila and the VASARI PON project.