Tourism Recommender Systems as a Vehicle for Social and Cultural Inclusion Fabio Gasparettia , Giuseppe Sansonettia and Alessandro Micarellia a Department of Engineering, Roma Tre University, Via della Vasca Navale 79, 00146 Rome, Italy Abstract Recommender systems for tourism have become so popular that our smartphones are now full of applications that can suggest customized itineraries anywhere and anytime. Most of them, however, recommend similar itineraries, usually even in the same overcrowded areas. In this article, we present the concept of an integrated framework for cultural tourism with different characteristics. Such a framework can propose alternative customized itineraries to favor cultural and social inclusion of visitors with local residents, for example, in urban suburbs or agricultural and industrial regions. Therefore, the system has to provide user interfaces to enable organizations, local enterprises, and visitors to analyze and exploit rich open data sources. In this way, local institutions could better plan and handle cultural tourism and public resources. Small businesses could cost-effectively promote their services. Visitors could receive personalized routes with knowledge related to local communities, cultures, traditions, and others. Keywords Recommender systems, Open data, Cultural heritage, Point of interest 1. Introduction and Motivations sibility of living satisfying experiences even following alternative routes in areas less overwhelmed by mass Tourism is increasingly one of the most relevant eco- tourism. In this way, it would act as an instrument of nomic sectors of a country1 . We are realizing this espe- social and cultural connection between the tourist and cially these days when the crisis due to the COVID-19 residents, as well as local service providers, producers, pandemic is seriously affecting all those who work in this and guides. Recent technological developments in net- sector. Among the different types of tourism, cultural working, Internet of Things, Artificial Intelligence, and tourism is one of the most important ways of revitaliz- Machine Learning [3] (e.g., Deep Learning [4]) play a ing the economy and stimulating other sectors related to central role in the creation of smart cities [5]. As a result, tourism [1]. The benefits of cultural tourism can be even many local institutions, regions, public and local organi- more important in the most disadvantaged areas. Cul- zations have begun to exploit the potential of open data tural tourism can, for example, help increase income and by providing and receiving access to rich repositories. employment, especially in disadvantaged social groups, Such data is often made available through open data ini- with fewer opportunities to enter the world of work: tiatives, such as the European Open Data Portal2 , thus young people, women, individuals with low education enabling the creation of services that combine data from and limited experience. In addition to this, urban suburbs, multiple sources. These services provide a wide range agricultural and industrial regions can represent destina- of data, including municipal information, data from ser- tions for an alternative cultural heritage, as long as they vices and companies, data from sensor networks, and can make themselves known as such. In this article, we meteorological information. One of the greatest benefits present the concept of an integrated system to promote of open data repositories is that they offer opportunities cultural tourism different from the usual one, in which for building personalized applications and services that tourists are often invited to visit the most popular places can enhance the user’s experience based on the current and to attend the most advertised businesses [2], thus context of use. As a result, local institutions and tourism also causing an overload of public services and utilities. service providers can direct visitors to places and services Our system should make the visitor aware of the pos- in underdeveloped areas, offering personalized itineraries enriched with interactive components and personalized Joint Proceedings of the ACM IUI 2021 Workshops, April 13-17, 2021, content. Overall, this system may ensure multiple bene- College Station, USA fits to different actors, such as the followings: Envelope-Open gaspare@dia.uniroma3.it (F. Gasparetti); gsansone@dia.uniroma3.it (G. Sansonetti); • tourists could receive suggestions of alternative micarel@dia.uniroma3.it (A. Micarelli) Orcid 0000-0003-4953-1390 (G. Sansonetti) cultural itineraries, personalized based on their in- © 2021 Copyright © 2021 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). terests, information related to their physical (e.g., CEUR CEUR Workshop Proceedings (CEUR-WS.org) Workshop Proceedings http://ceur-ws.org ISSN 1613-0073 location, weather conditions, means of transport, 1 https://www.tourism-review.com/ 2 tourism-industry-is-the-pillar-of-economy-news11210 https://www.europeandataportal.eu/en time of the day, day of the week) and social (e.g., for some brilliant solutions of specific aspects. For ex- social media network) context [6], and open data. ample, the PersTour system [13] takes into account the The system could also enable tourists to express user’s interests and the popularity of POIs, also indicat- their feedback on the experienced itinerary and ing the time to be spent in each of them. In general, POIs to share their impressions through social media; represent particular places. However, sometimes POIs • tourists could reap significant benefits from a sys- can represent larger areas (for example, entire architec- tem capable of suggesting alternative cultural her- tural or cultural districts, or street markets) in which itage not only in terms of perceived accuracy, but tourists may want to take walks, such as in [14]. We also of novelty, serendipity, and diversity of the propose a system that can recommend complete person- recommendation list [7]; alized experiences to the target user. Specifically, our • local administrations could better plan and man- system must be able to provide the user, and possibly age the impact of tourism on public services and her travel companions as well, with recommendations utilities such as local transport. Furthermore, the of POIs and itineraries among them. For this purpose, system could suggest alternative itineraries to the advanced user modeling techniques are required that al- traditional ones, normally concentrated on a lim- low the system to select suitable textual and multimedia ited number of well-known sites, thus mitigating content [15]. Those techniques must be able to exploit the overcrowding of the most visited areas; heterogeneous information such as social data (obtained • tourists, local administrations, small businesses, by analyzing the user’s activity on social media), sensor service providers, and residents could collabora- data, and open data [16]. To the best of our knowledge, tively create itineraries, thus fostering intercul- there are no tourist recommendation systems in the liter- tural and social exchange through greater aware- ature capable of suggesting personalized experiences to ness of different cultural identities acquired by the user based on the integration of all such data. In [17], sharing traditions, customs, cuisine, and other the authors propose an approach to take advantage of information on cultural tourism; linked open data (LOD) and generate a recommendation • all the actors involved could participate in the pro- of personalized itineraries with related textual and mul- motion of alternative cultural heritage through timedia content. The recommendation engine considers a curation process, according to the models and the active user profile, the current context of use, and methods developed, for example, as part of the the POIs extracted from LOD. Novel aspects of the pro- European H2020 SPICE project3 . posed system are the extraction and subsequent filtering of POIs through dynamic queries as well as the definition of the itinerary taking into account the popularity and 2. Idea and Background diversity of the POIs in terms of their categories, the over- all length of the itinerary, and the travel time. A similar The problem of personalizing the tourist experience con- approach is illustrated in [18], where the author proposes sists in determining a plan with a sequence of visits the personalized recommendation of individual POIs us- of a given number and categories of points of interest ing the information extracted from social networks such (POIs) [8]. For instance, a tourist might be interested as Facebook4 , and the target user’s preferences and in- in having lunch in a local cuisine restaurant, taking an terests captured asking her to rate a sample of chosen educational tour with some stops of cultural interest, and images reflecting particular categories of POIs. Another ending the day in a popular music club. Another tourist noteworthy recommender is the one proposed in [19], may want to eat some street food and then go to a local which first identifies and assigns a relevance score to spe- folklore show. These sequences of visits must be carried cific POIs, then suggests routes of interest among them. out in a predetermined time interval and must be accom- However, unlike the recommender we propose, which panied by information for each visit to a POI, the total relies on open data, the authors employ a closed database cost of the itinerary, and the orientation on the map. In (i.e., Foursquare5 ) as a data source for their system. In order for visitors to receive personalized visit plans, a rec- developing our recommender, we intend to leverage an ommender system (RS) is needed. RSs are software tools open-source dashboard technology and data manage- that provide the target user with suggestions of items ment technology that gives a single point of access to likely to be of her interest [9]. They are successfully ap- open data. For this purpose, several integrated platforms plied in manifold domains, including music, movies, and are available, including Digital Enabler6 , Snap4City7 , and research papers [10]. Several RSs for tourists have also been proposed [11, 12]. More specifically, many systems can recommend POIs and itineraries among them. Some 4 https://www.facebook.com/ 5 recent itinerary recommenders deserve to be mentioned https://foursquare.com/ 6 https://digitalenabler.eng.it/suite/ 3 7 https://spice-h2020.eu/ https://www.snap4city.org/ OneSait8 . Those platforms provide tools for developing [6] D. D’Agostino, F. Gasparetti, A. Micarelli, G. San- the Internet of Things and managing data collected from sonetti, A social context-aware recommender multiple sources. They also furnish dashboards for view- of itineraries between relevant points of interest, ing data by users as well as local institutions and tourism- in: C. Stephanidis (Ed.), HCI International 2016 – related organizations, thus enabling all of them to create Posters’ Extended Abstracts, volume 618, Springer itineraries in a collaborative way. International Publishing, Cham, 2016, pp. 354–359. [7] M. Kaminskas, D. Bridge, Diversity, serendipity, novelty, and coverage: A survey and empirical anal- 3. Conclusions ysis of beyond-accuracy objectives in recommender systems, ACM Trans. Interact. Intell. Syst. 7 (2016). In this article, we have introduced the concept of an [8] D. Gavalas, C. Konstantopoulos, K. Mastakas, integrated system for first collaboratively creating and G. Pantziou, Mobile recommender systems in then recommending alternative personalized itineraries tourism, Journal of Network and Computer Ap- to tourists in areas other than those normally visited plications 39 (2014) 319–333. by mass tourism. Personalization takes place based on [9] F. Gasparetti, G. Sansonetti, A. Micarelli, Commu- the user’s interests, the physical and social context of nity detection in social recommender systems: a use, and nearby businesses and services. This system re- survey, Applied Intelligence (2020). lies on open data and allows local institutions and small [10] H. A. M. Hassan, G. Sansonetti, F. Gasparetti, A. Mi- businesses to access it via a dashboard. In this way, in- carelli, Semantic-based tag recommendation in sci- stitutions can better plan and manage public services entific bookmarking systems, in: Proceedings of and utilities, whilst small businesses can make visitors the 12th ACM Conference on Recommender Sys- aware of their offers without having to invest financial tems, ACM, New York, NY, USA, 2018, pp. 465–469. resources to promote themselves. The goal of our system [11] K. Chaudhari, A. Thakkar, A comprehensive sur- is to offer the visitor personalized itineraries in underde- vey on travel recommender systems, Archives of veloped or disadvantaged areas, such as urban suburbs or Computational Methods in Engineering 27 (2019). agricultural and industrial regions, highlighting them as [12] G. Sansonetti, F. Gasparetti, A. Micarelli, F. Cena, alternative destinations for a new cultural tourism. This C. Gena, Enhancing cultural recommendations way, the system could promote and favor the inclusion through social and linked open data, User Modeling of tourists in the social and cultural fabric of the places and User-Adapted Interaction 29 (2019) 121–159. they are visiting, thus enhancing social development and [13] K. H. Lim, J. Chan, C. Leckie, S. Karunasekera, Per- a greater intercultural awareness of values and cultural sonalized trip recommendation for tourists based on identities by exploring new different forms of cultural user interests, points of interest visit durations and heritage. visit recency, Knowl. Inf. Syst. 54 (2018) 375–406. [14] D. Gavalas, V. Kasapakis, C. Konstantopoulos, References G. Pantziou, N. Vathis, Scenic route planning for tourists, Personal Ubiquitous Comput. 21 (2017). [1] M. Ursache, Tourism - significant driver shaping a [15] G. Sansonetti, F. Gasparetti, A. Micarelli, Cross- destinations heritage, Procedia - Social and Behav- domain recommendation for enhancing cultural ioral Sciences 188 (2015) 130–137. heritage experience, in: Adjunct Publication of [2] G. Richards, Cultural tourism: A review of recent UMAP’19, ACM, New York, NY, USA, 2019, p. research and trends, Journal of Hospitality and 413–415. Tourism Management 36 (2018). [16] A. De Angelis, F. Gasparetti, A. Micarelli, G. San- [3] L. Vaccaro, G. Sansonetti, A. Micarelli, An empirical sonetti, A social cultural recommender based on review of automated machine learning, Computers linked open data, in: Adjunct Publication of UMAP 10 (2021). ’17, ACM, New York, NY, USA, 2017, pp. 329–332. [4] G. Sansonetti, F. Gasparetti, G. D’Aniello, A. Mi- [17] A. Fogli, G. Sansonetti, Exploiting semantics for carelli, Unreliable users detection in social media: context-aware itinerary recommendation, Personal Deep learning techniques for automatic detection, and Ubiquitous Computing 23 (2019) 215–231. IEEE Access 8 (2020) 213154–213167. [18] G. Sansonetti, Point of interest recommendation [5] G. D’Aniello, M. Gaeta, F. Orciuoli, G. Sansonetti, based on social and linked open data, Personal and F. Sorgente, Knowledge-based smart city service Ubiquitous Computing 23 (2019) 199–214. system, Electronics (Switzerland) 9 (2020) 1–22. [19] W. Wörndl, A. Hefele, D. Herzog, Recommending a sequence of interesting places for tourist trips, Information Technology & Tourism 17 (2017) 31–54. 8 https://onesaitplatform.atlassian.net/