Proceedings of the 22th International Conference on Automation & Computing, University of Essex, Colchester CO4 3SQ, 7-8 September 2016
Overview of Personalized Travel Recommendation Systems Pree Thiengburanathum(PhD student)a, Shuang Cangb, Hongnian Yuc, Faculty of Science and Technology, Bournemouth University, UK. Email:
[email protected],
[email protected] Faculty of Management, Bournemouth University, UK. Email:
[email protected] Abstract— The number of tourists worldwide has increased rapidly. The extensive amount of information that is accessible on the Internet provides tourists with a lot of choices. However it is a difficult task for the tourists to select the right tourism products or services based on their preferences and requirements. This paper attempts to review the published studies on Travel Recommendation System (TRSs) during 2008 to 2015. The paper emphasizes the use of previous and current Information Communication Technology (ICT), its applications, possible research trends and the challenges that arise in the development of a TRS.
information searches in the future will respond to travelers’ concerns when planning trips, booking reservations, and purchasing tickets [4]. Decision support tools, are known as Recommendation Systems (RSs), have been developed to address these concerns. In the tourism field, they are referred to as Tourism Recommendation System (TRS). Not only can a TRS be used to assist tourists and tourism providers in the process of selecting products and services in an unfamiliar city, but it can also help promote tourism in with that city.
Keywords: Personalized Recommendation Systems; ETourism; Review
I.
INTRODUCTION
Travel and Tourism are extremely important globally, contributing 10% to the world economy in 2015, and it is projected to grow to an estimated 10.3% in this decade [1]. Travel and Tourism has substantially benefited from ICT, especially in terms of Internet technology [2]. The internet has become the main information source for tourists over the last decade, playing and important role in helping tourists choose produces and services. Consequently the sheer volume of data available on the Internet can overwhelm tourists and travel agents, making it difficult to process the information, whether they are planning their trip, making their journey, or after their travels [3]. Today, tourists and tourism providers can search, compare, choose, and make decision almost instantly. In 2014, thirty per cent of reservations were made online, a number that is expected to double in the next five years [2]. The nature of a travel search is complex and a dynamic problem in that there are many factors involved in the decision-making process; for example, the number of nights one may choose to stay, the price of meals, the budgets of the travelers, hotels, the number of travelers, time, things to do, etc. In order to be competitive and profitable and to make life easier for tourists, tourism industries and travel agencies need to make use of ICT to ensure the best services to tourists. The greater tourism industry, travel agencies and tourism companies require ICT to deliver quality services and to remain competitive. Furthermore, online
Fig 1. Increasing the use of ICT in tourist services searches.
The main objective of this paper is to investigate whether there are new technologies, trends or challenges involving the TRS. This paper has reviewed the TRS developed during the period of 2008 to 2015. The published sources used in this research study were selected from well-known online libraries such as Sciencedirect and Google scholar, and citations from articles published after 2008. The keywords, ‘travel, trips, tourism, recommendation systems’ were used. Thirtythree journal articles form our search were obtained. At this stage, this paper aims to clarify the state of the art in ICT as it has emerged in TRS development. In addition, the TRS applications which have the greatest potential to contribute to the overall body of tourism knowledge in terms of both academic and practical impacts were identified. This paper has the following sections. Section 2 provides the relevant background of a TRS. Section 3 presents the available ICT that has been used in the recent TRS. Section 4 explains the tourism services provided by the RS. The challenges and future trends of a personalized TRS will be presented in Section 5.
II.
BACKGROUND
A. Travel Recommendation Systems Travel and Tourism activities involve complex decision-making processes; for example, the process of selecting destinations, attractions, activities, hotels, restaurants, and services by the tourist or tourism agent. Thus, many academic and industry researchers are interested in TRSs. Over the past six years, most TRS studies have appeared in the Expert Systems with Applications journal (See Table 1) They have been developed/deployed across many platforms; for examples, desktop applications, web applications, and tablet/mobile applications. Based on user input, TRSs may: 1) recommend results that are based on estimations of user interest; 2) recommend Points of Interests (POIs), tourism services, or routes; 3) rank suggested attractions/destinations in sequence; or 4) propose a holistic trip plan. Initially, TRSs have focused on recommending destinations, along with integrating certain tourism services, such as hotels and restaurants, into the content. The output of most systems is itinerary-based. Lately, researchers have expanded their focus on recommending a route and toward solving trip/itinerary design problems. Many TRSs provide a holistic trip plan by mainly focusing on specific content. One of the main tasks for the recommendation engine is to classify or to cluster the items. For example, match the right item(s) to the user(s). Therefore, many measures of established similarity methods have been applied in the previous TRSs. It can be seen that the easiest and most common method is the Euclidean distance; which used the Euclidean distance between each pair of user and the designated activity [5]. Cosine Similarity, also known as the L2Norm is another common method that can be used to determine the similarity between users [6]. And finally, Pearson’s correlation from statistics has also been adapted to use in TRS to find the similarity between users/items (i.e., linear relationship between two sets of data) [7]. Previous TRSs rely heavily on knowledge-based recommendation techniques (i.e., both case-based and constraint-based). It can be seen that they have moved from traditional recommendation approaches (i.e. collaborative, content-based, and knowledge-based) to context-based recommendation and hybrid-based recommendation approaches. The concept of context in RS can be in the field of Information Retrieval(IR), ubiquitous and mobile context-aware systems, marketing, and management [8]. TRSs that are implemented with a context-based approach rely on a network of sensors to collect contextual information and they are mostly a pullbased system (i.e. demanding human intervention) [9]. PSiS [10] is a mobile TRS that provides POIs
recommendation focusing on user sight context (e.g., location, time, speed, direction, weather) and user preferences (e.g., through their previous work). The system has the capability to adapt dynamically to a recommended tour; for example, it can re-generate a new trip plan when the user is ahead of schedule. Another interesting feature is the architectonic tag, which can recommend POIs according to whether a destination is open or closed and is worth the visit. An additional feature is the tracking system, which can track the tourist during the trip, with the benefit of saving time. TABLE I.
PUBLICATION OF TRAVEL RECOMMENDATION SYSTEM FROM 2008-2015 Number of publication
Impact factor(2014)
Expert Systems with Applications
18
2.24
Decision Support Systems
1
2.313
Information Fusion
1
3.681
IET software
2
0.658
Telematics and Information
1
1.12
Sensor
1
1.852
Applied Soft Computing
1
2.14
1
1.467
1
1.625
Tourism Management
1
2.554
Journal of Tourism
1
n.d
1
1.518
1
0.321
1
1.485
Journal of Web Semantics
1
2.55
Total
33
Journal
Journal of Network and Computer Applications Engineering Applications of Artificial Intelligence
Personal and Ubiquitous Computing International Journal on Artificial Intelligence Tools Journal of Systems and Software
III.
TECHNOLOGIES INVOLVED IN E-TOURISM
Fig 2. Emerging of ICT
Judging by the post 2008 TRSs, most of the TRSs rely heavily on the hardware, software, and communication technologies (See Figure 2). In this section, the ICT aspects that have been adapted in the TRS development process since 2008 are discussed.
A. Wireless Sensor Networks Recently, researchers have studied the effects of mobile and wireless technologies, including mobile telephones, and wireless data communication on TRS. These technologies enhance the recommendation systems for tourists in terms of context-awareness, real-time recommendations, opportunities to redesign the route during the trip, and adapting to changed circumstances, as can be seen in [9], [11]–[13]. The Global Positioning System (GPS) and Geographic Information System (GIS) are used to retrieve user locations, provide user directions, detect nearby friends, calculate travel speed, and detect nearby Points of Interest (POIs). This helps the user find the best POIs or routes, both before and during travel. Many TRSs are not only deployed as stand-alone applications on desktops or as browser platforms, but they are also supported on mobile devices due to the prevalence of smart phones embedded with GPS, compasses, accelerometers, and other sensors. With mobile applications, parameters such as weather, noise levels, and people nearby can be considered in the recommendations. Also, 3G, 4G, Wi-Fi, WiMAX, and Bluetooth communication networks provide researchers with more opportunities and new state-of-the-art resources. Wireless technology has been used in recent TRSs. For instance, Tsai [14] proposes a personalized TRS for theme parks helps tourists select a ride based on real time information collected by radio-frequency identification (RFID). Gavalas [15] implementes a Mobile Tourism Recommendation System (MTRS) that deploys the Wireless Sensor Networking (WSN) infrastructure to solve the problem of delivering a cost-effective means for remote content updates and to support proximity detection [16]. There are two challenges and one innovation for this TRS, referred to as a Mobile Tourism Recommendation System (MTRS). Firstly, there is the use of context-aware rating as a collaborative filtering approach in MTRS where tourists can upload, review and make comments via their mobile devices. Secondly, there is an attempt to implement a wireless sensor-networking (WSN) infrastructure to solve the problem of providing a cost-effective means for remote content updates and to support proximity detection (rural positioning of points of interest). The input data come from the user registration from websites, where the input variables involve gender, marital status, age, education level, POI categories, and favourite leisure activities as optional in out category. WSN is the innovation of this research. Due to the lack of a developed network infrastructure and the high cost of mobile services in many countries, tourists mostly avoid using 3G/Edge connections[17]. However this TRS still suffers from the implementation of unreliable networks. The Internet of Things (IoT) is another concept that may play an important role in the tourism industry.
According to [18], IoT refers to the trend to merge the physical world with the world of information in a general Internet-like state of connectedness. For example, IoT connects many objects, stakeholders, agents, and subsystems in the business process. Therefore, tourists can now generate, send and receive data through communication devices via a range of communication technologies, networking protocols and data types, with little human intervention. B. Artificial Intelligence Artificial Intelligence (AI) has been applied to tourism research lately. AI has many different definitions, but, to put it simply, it is the technology that seeks to understand the human thought process and to simulate human intelligence in machines [19].
Fig. 3 Bayesian Network model to predict a tourist’s favorite attraction [17]
AI and Machine learning have been heavily adapted in TRS to improve the decision-making process, optimization, scheduling, clustering, knowledge representation, and planning. Figure 3 shows that the Bayesian Network (BN), sometimes known as a belief network or probabilistic directed acyclic graphical model, is one of the popular machine learning techniques that TRS researchers use to estimate user preferences. A BN combines Bayesian theories about knowledge. For example, given demographic tourist information, a BN estimates a tourist’s preferred attraction or activity [7], [20]. A BN is a hybrid recommendation system that combines content-based filtering and collaborative filtering [20], [21]. Fuzzy logic has also been adapted in previous studies, mostly for knowledge based TRSs [22]. The fuzzy method has been used to deal with the uncertainties that surround linguistic assessments taken from sector experts and tourist feedback [23], [24]. The fuzzy system has also been used to learn the uncertainty of driver behaviour to make the recommendation system more intelligent, by understanding the imprecise (fuzzy) way a driver picks a route [25]. Case-Based Reasoning (CBR), a machine learning method, provides solutions to similar problems involving four processes: retrieve, reuse, revise and retain. Multiple-Criteria Decision Analysis (MCDM), another problem-solving mythology, is a good method to evaluate and compare criteria and then rank the alternatives. Alptekin and Buyukozkan [23] proposed an intelligent tourism destination planning system to help travel
agencies reduce their workload. The system combines CBR and MCDM to increase system accuracy, where both of the methods share something in common in terms of decision-making. The challenges of this research study involve the integrating of these two decision-making methods and an understanding of how to increase the accuracy of the TRS. User requirements such as tour type (e.g. active, wondering, city), number of travelers, region, transport mode, tour length, season, accommodation type, and rating (i.e. number of stars) are the parameters for the TRS. The output of this TRS is the travel plan with a quoted price. The advantages of the system are that the reliability of the obtained result and the framework can be adapted to other application domains. A major disadvantage of this system is the adaptation feature, which relies heavily on the experiences of the travel agencies. For example, when a tourist creates a new case, the new case cannot be inserted right into the database, as it has to be evaluated by the travel agency or accepted by the tourist first. Another disadvantage is the cold-start problem; because this TRS requires a large amount of time to collect data and convey it to the database. A Genetic Algorithm (GA) is a search heuristic that mimics the process of natural evolution. Ant Colony Optimization (ACO) is a meta-heuristic method that mimics ant behavior. Both have also been used by personalized TRS to learn about tourist personalities and context data in order to select a suitable route or POIs for them [12], [26], [27]. C. Ontology and Sematic Web technology The goal of the Sematic Web, also known as Web 3.0, is to efficiently share data process information automatically and manually by promoting common exchange protocols and data formats. Many TRSs rely heavily on knowledge in the tourism domain. In order to represent the knowledge, the technology called ontology has been commonly used. Ontology is a method used in Computer Science and Information Sciences. It helps represent knowledge in the domain, or at least part of it, as a set of concepts. It considers the relationship between the knowledge and also plays a prominent role in the framework in a sematic web[20]. SAMAP [28] is one of the examples of TRSs that has been successfully modeled and implemented. Their ontology represents tourist’s interest (e.g. user, city, transport, place, personal preferences). Sematic Web technology and Ontology helps the researcher integrate the heterogeneous online information[12], [13], [20], [28]–[31]. Resource Description Framework (RDF) and Web Ontology Language (OWL), are the most commonly used languages [29] and have been used to develop TRSs that represent classes and their relationships. SPETA [11] takes advantages of Web 3.0 technologies by integrating social networks, the sematic web, and context-awareness as a mobile TRS. The system
aims to recommend tourism services such as attractions or restaurants to the tourist. The TRS is focused on matching, searching, and filtering items from the knowledge acquired on ontology (i.e. social and geolocation information). The system requires input from the user - both explicitly and implicitly – in order to make a recommendation. The input includes user preferences (food and music type), user context information (weather, time, location), and derived variables such as speed and direction. The system also incorporates the opening and closing times/dates of attractions. D. Agent technology Agent technology provides many benefits in modeling complex real-world problems. Many personalized TRSs have adapted this technology to their research [5], [28], [32]. Multi-agent systems (MAS) are composed of agents that interact with each other in the environment. Each agent has its own goal and tries to maximize resources, utilization and benefits [33]. MAS is as promising tool for modeling problems of organization or real world problems where people have to make decisions as a group [34]. Some agents in the system are identified as Intelligent Agents (IA), since they can make decisions, optimize, schedule, and solve complex problems. Turist@[5] is one example of a TRS that has been implemented with MAS. It is a mobile push and locationbased TRS that has a high degree of dynamic adaptability, taking user GPS locations into account (i.e., the system can adapt to changes in the trip schedule and incorporate new suggestions). The system also considers user demographic information (e.g., date of birth, education, nationality, language, disability), trip characteristics (travel group type, trip duration) and user preferences. The system notifies the user when she/he is near an activity and suggests interest activities of interest. The TRS uses the hybrid filtering method combining content-based filtering and collaborative filtering to give a personalized recommendation. The TRS has a feature that can provide the dynamic management of the user profile to be used in the personalized recommendation process, such that the profile will be updated in both explicit and implicit ways. The use of a MAS provides many advantages for the distributed system, such as there is an agent run on the mobile device, a broker agent running as the facilitator between the user agent and the activity agent handling communication between them, and another agent is responsible for the maintenance of the databases so as to reduce the server overload and so on. Moreover, the ability to adapt, adjust, add and remove the agents seems to be the concept for modularity design when modeling a distributed system and real world problems. Also, there is a high degree of adaptive capability of the system, such as the system could adjust to a new plan based on the new location of the user at the time of execution. The user feedback is based on both explicit and implicit functions, for example, the rating approach for the former, and with
regard the latter, the act of monitoring user action by analyzing the time the user spend on the webpage or the link that the user followed, etc. E. Web design When tourists browse travel websites, they expect the websites to be adaptive, interactive, responsive, provide dynamic information, and to be attractive [35]. To meet this expectation, many personalized TRSs have used Ajax and other advanced web programming and powerful frameworks which combine several technologies, such as HTML, JavaScript, XML, and document object model, to create a sense of interaction between the user and the web application [35]. Chiang [36] proposed a travel planning system for a personalized travel schedule that has an adjustable interface module, which could enhance travel planning flexibility. Moreno [38] developed a web-based TRS by using JSF and Ajax. The ontology has been developed using the thesaurus of the WTO as reference guide with the OWL. Buhalis [39] claims that web design was one of the most important technology innovations for the tourism industry. Moreover, accessibility features for disabled and elderly people should be considered as a beneficial feature of an interactive website. IV.
E-TOURISM SERVICES FROM RECOMMENDATION SYSTEM
Fig. 4 Applications in TRS
Major TRSs have focused on recommending destinations, along with integrating tourism services, such as hotels, and restaurants, into the content as is shown in Figure 4. The output of most systems is itinerary- based. Lately, researchers have expanded their focus on recommending a route toward to addressing trip/itinerary design problems. Many TRSs provide a holistic trip plan by mainly focused on specific content. From the relevant literature, TRS can be categorized based on the e-tourism services they provide, as follows: A. Destination and tourist services recommednation In the simplest terms, Destination TRSs or DRSs list destination such as POIs, attractions, activities, events according to specific input constraints from users. Some of them take context information into account. DRSs are moving toward the stage of being able to rank the importance of destinations and predict destinations
suitable for the user [40]. Some DRSs have used decision-making theories to better understand how a tourist selects preferred destinations and to improve recommendation accuracy [7], [20]. B. Route recommendation Wireless Sensor Network technologies, like GPS and RFID, can retrieve context information such as current location, as a parameter. A Route TRS can recommend route(s) through several destinations for a tourist. For example, it can learn user behavior through context information to predict a route based on a user or a group’s preferences [12], [14]. Route TRSs provide point-to-point recommendations with multi-model transportation services [27], [28]. Additionally, there is a TRS that provides real-time information to tourists to reduce congestion and avoid long queues at tourist hotspots [26]. C. Trip planning/ itinerary recommendation Trip planning is challenging; for example, tourists usually have specific requirements and needs, such as the number of nights to stay, the number of travelers, budget, required destinations, the days attractions are open, and starting locations. Trip planning/itinerary recommendation systems take these user preferences and/or context features into account in deriving the order of destinations for an itinerary. Moreover, these systems can create a new plan/itinerary for a traveler in response to changes incurred during the trip. For example, if the traveler is running out of time, the system may reschedule a destination. While TRSs cover many different aspects of tourism services, few focus on the trip planning or scheduling problem, as this is a complex problem that requires the TRS to generate an automated optimal travel plan (i.e., the most realistic travel plan) for the user based on many constraints. According to [41], this problem has been termed a tourist itinerary design problem (TIDP) or tourist trip design problem (TTDP) [15], [42], [17], [42]. This problem resembles the classic Travelling Salesman Problem (TSP) in theoretical computer science and operations research. However, the TSP conundrum is concerned with minimizing travel time or travel distance. The simplest TIDP can be modeled as an Orienteering Problem (OP) where a set of vertices is given points of interest, each of them having a score (e.g. user satisfaction) and the goal is to create the best path to maximize the total score (time or budget) from each of the vertices. Golden [43] proved that OP is a NP-hard problem. TIDP can be modeled as a team orienteering problem (TOP), where the problem is NP-complete [44]. The team orienteering problem with time windows (TOPTW) (e.g. considering the opening and closing times per day), which has appeared in recent studies [17], [45], is an extension of the TOP.
V.
FUTURE RESEARCH TRENDS AND CHALLANGES
A)
User constraints and contextual information for realistic trip plan Recommending a near optimal or realistic trip itinerary is a major challenge, such that the following user constraints and context constraints can be added to the TRS to generate more realistic and effective recommended trip plans. This is done to satisfy user requirements and his/her preferences [45]–[47]. The following user constraints and contextual information can be added to the TIDP model. The City Trip Planner [45] assists a tourist when planning routes for five cities in Belgium. The system aims to deal with the TOPTW problem with the trip planning heuristic algorithm. In addition to trip constraints including number of days, starting and ending locations, starting and ending times, lunch breaks, and opening and closing times, the system weighs user preferences to estimate interest in each POI. PTPS [36] is a web-based TRS that recommends and schedules hotels, restaurants, and attractions for the user, considering all categories of user requirements (e.g. number of days, number of travelers, budgets, lunch and dinner times, required POIs, and starting point). The proposed system recommends POIs based heavily on user needs/requirements in order to achieve maximum user satisfaction. The system also introduced an algorithm to solve the TIDP. Moreover, with the adjustable interface feature embedded in the system, users can adjust their results to replace unsatisfying items and to improve suggestions. There is room for more research on constraint-based and context-based recommendation systems, not only in the tourism domain but in the applications including map navigation, fleet management, weather information, roadside assistance, and personal location services [9], [12]. B)
User constraints and contextual information for destination selection TRSs provide options in selecting destinations and services by taking into consideration the users’ hard constraints (e.g., contextual information, requirements, preferences, interests, and demographics) and destination information. Future TRSs should provide a traveler with even more options (soft constraints) to force the system to collect information on the destination that he/she wants to visit based on his/her needs. For instance, some tourists do not want to visit more than a specific number of destinations per day or the destination that he/she has already visited on a previous trip[47] Since most users are budget-conscious, the travel budget should include limits for transportation fees, event entrance/admission fees, and hotel/restaurant bills. Moreover, the number of travel days, accessibility (hearing, motor disabilities, and blindness) should be taken into account [47].
C) User constraints for tourist services selection Soft constraints can be added to a TRS. For example, a TRS that recommends restaurants could be programmed to incorporate meal times, food type (Chinese, Thai, etc.) and price range (low-high). With these soft inputs, the TRS could recommend restaurants with opening hours and a price range that match the user’s selection criteria. For a TRS that recommends hotels, soft constraints can also be added, such as hotel type, price range, and amenities [47]Transportation options should be based on a multi-option model (i.e., travelers can first take a taxi, then walk to the POI) and certain other aspects regarding transport services (e.g., transport fee) [17], [28]. Regarding the contextual information, weather, traffic forecasting, and current date/time to match the destination operation date/time should be taken into consideration [47]. D) Integration of heterogeneous online travel information Integrating heterogeneous online travel information is a major challenge for TRSs [20]. TRSs involve gathering large amounts of information from different information providers or tourism services (e.g. hotels, restaurants, POIs) with different, or even unique, types of categories or content in a variety of formats including any nonstructural data. To address this challenge, information extraction techniques such as web extraction/crawler [37], semantic technologies, and Web 2.0 technologies, such as Mashup, a content aggregation technology [5], [20], [28], have been adopted by TRS researchers. A normal Relational Database Management System (RDBMS) would have difficulty managing the large amount and complex nature of data used in TRSs, including geo-spatial data and continual and numerous user updates, given data availability and scalability issues. For TRSs, Not Only SQL (NoSQL) is a more promising technology for increasing system performance and reducing latency than RDBMSs. As a trade-off of using NoSQL, TRSs may lose database-wide or transaction consistency[52]. E) Group-based recommendation Group-based recommendation systems pose a challenge because not only do group tourists have different individual preferences, but they also must be concerned with the preferences of other group members. Recommending an itinerary for a group that optimally satisfies the differing individual interests is difficult. Given this difficulty, only one TRS study attempts to support both individual and group travelers [49]. VI.
CONCLUSION
This paper has focused on reviewing the ICT used in the TRS development, e-tourism services that TRS currently provide, and recent trends/challenges in the TRS research scene. It can be seen that the latest ICT provides the new opportunity for researchers to design and develop a TRS to become more intelligent, interactive, adaptive, automatable, as well as to support
high degree of user satisfaction and user experience than ever before. Furthermore, contextual information, user requirements, interests, preferences, and sociodemographic information have been taken into account as factors when designing a TRS. In addition, new generation TRSs display the ability to adapt to the use of contextual information features that let the user modify the result, along with feedback mechanisms that can improve recommendation accuracy rate, user experience, and user satisfaction. However, there are still more aspects in TRS development that we need to investigate, such as theories to improve the level of personalization, methodologies, system evaluations, and other inherent challenges found within the TRS realm.
[11]
[12]
[13]
REFERENCES [1]
World travel and tourism council, “Travel and Tourism enconomic impact 2015 Thailand.” [2] E. Pitoska, “E-Tourism: The Use of Internet and Information and Communication Technologies in Tourism: The Case of Hotel Units in Peripheral Areas,” Tourism in South East Europe, vol. 2, pp. 335–344, Dec. 2013. [3] E. Pantano and L. D. Pietro, “From e-tourism to ftourism: emerging issues from negative tourists' online reviews,” Journal of Hospitality and Tourism Technology, vol. 4, no. 3, pp. 211–227, 2013. [4] S. (Shawn) Jang, “The Past, Present, and Future Research of Online Information Search,” Journal of Travel & Tourism Marketing, vol. 17, no. 2–3, pp. 41–47, 2004. [5] M. Batet, A. Moreno, D. Sanchez, D. Isern, and A. Valls, “Turist@: Agent-based personalised recommendation of tourist activities,” EXPERT SYSTEMS WITH APPLICATIONS, vol. 39, pp. 7319–7329, 2012. [6] S. Schiaffino and A. Amandi, “Building an expert travel agent as a software agent,” Expert Systems with Applications, vol. 36, no. 2, Part 1, pp. 1291– 1299, Mar. 2009. [7] F. M. Hsu, Y. T. Lin, and T. K. Ho, “Design and implementation of an intelligent recommendation system for tourist attractions: The integration of EBM model, Bayesian network and Google Maps,” EXPERT SYSTEMS WITH APPLICATIONS, vol. 39, pp. 3257–3264, 2012. [8] F. Ricci, L. Rokach, and B. Shapira, “Introduction to Recommender Systems Handbook,” in Recommender Systems Handbook, F. Ricci, L. Rokach, B. Shapira, and P. B. Kantor, Eds. Springer US, 2011, pp. 1–35. [9] C. Lamsfus, D. Martin, A. Alzua-Sorzabal, D. López-de-Ipiña, and E. Torres-Manzanera, “Context-based tourism information filtering with a semantic rule engine,” Sensors (Basel, Switzerland), vol. 12, no. 5, pp. 5273–5289, 2012. [10] R. Anacleto, L. Figueiredo, A. Almeida, and P. Novais, “Mobile application to provide personalized sightseeing tours,” Journal of
[14]
[15]
[16]
[17]
[18]
[19] [20]
[21]
[22]
[23]
Network and Computer Applications, vol. 41, pp. 56–64, May 2014. A. García-Crespo, J. Chamizo, I. Rivera, M. Mencke, R. Colomo-Palacios, and J. M. GómezBerbís, “SPETA: Social pervasive e-Tourism advisor,” Telematics and Informatics, vol. 26, no. 3, pp. 306–315, Aug. 2009. J. A. Mocholi, J. Jaen, K. Krynicki, A. Catala, A. Picón, and A. Cadenas, “Learning semanticallyannotated routes for context-aware recommendations on map navigation systems,” Applied Soft Computing, vol. 12, no. 9, pp. 3088– 3098, Sep. 2012. F. M. Santiago, F. A. López, A. Montejo-Ráez, and A. U. López, “GeOasis: A knowledge-based georeferenced tourist assistant,” Expert Systems with Applications, vol. 39, no. 14, pp. 11737–11745, Oct. 2012. C.-Y. Tsai and S.-H. Chung, “A personalized route recommendation service for theme parks using RFID information and tourist behavior,” Decision Support Systems, vol. 52, pp. 514–527, 2012. D. Gavalas, M. Kenteris, C. Konstantopoulos, and G. Pantziou, “Web application for recommending personalised mobile tourist routes,” IET Software, vol. 6, pp. 313–322, 2012. D. Gavalas and M. Kenteris, “A web-based pervasive recommendation system for mobile tourist guides,” PERSONAL AND UBIQUITOUS COMPUTING, vol. 15, pp. 759–770, 2011. D. Gavalas, M. Kenteris, C. Konstantopoulos, and G. Pantziou, “Web application for recommending personalised mobile tourist routes,” IET Software, vol. 6, no. 4, pp. 313–322, 2012. Melanie Swan, “Sensor Mania! The Internet of Things, Wearable Computing, Objective Metrics, and the Quantified Self 2.0,” Journal of Sensor and Actuator Networks, no. 3, p. 217, 2012. E. Turban, R. Sharda, and D. Delen, Business Intelligence and Analytics: Systems for Decision Support, 10 edition. Harlow: Pearson, 2014. Y. X. Huang and L. Bian, “A Bayesian network and analytic hierarchy process based personalized recommendations for tourist attractions over the Internet,” EXPERT SYSTEMS WITH APPLICATIONS, vol. 36, pp. 933–943, 2009. F. Sparacino, “Sto(ry)chastics: A Bayesian Network Architecture for User Modeling and Computational Storytelling for Interactive Spaces,” in UbiComp 2003: Ubiquitous Computing, vol. 2864, A. Dey, A. Schmidt, and J. McCarthy, Eds. Springer Berlin Heidelberg, 2003, pp. 54–72. J. P. Lucas, N. Luz, M. N. Moreno, R. Anacleto, A. Almeida Figueiredo, and C. Martins, “A hybrid recommendation approach for a tourism system,” Expert Systems with Applications, vol. 40, no. 9, pp. 3532–3550, Jul. 2013. G. I. Alptekin and G. Buyukozkan, “An integrated case-based reasoning and MCDM system for Web based tourism destination planning,” EXPERT
[24]
[25]
[26]
[27]
[28]
[29] [30]
[31]
[32]
[33] [34] [35]
[36] [37]
SYSTEMS WITH APPLICATIONS, vol. 38, pp. 2125–2132, 2011. A. Garcia-Crespo, J. L. Lopez-Cuadrado, R. Colomo-Palacios, I. Gonzalez-Carrasco, and B. Ruiz-Mezcua, “Sem-Fit: A semantic based expert system to provide recommendations in the tourism domain,” EXPERT SYSTEMS WITH APPLICATIONS, vol. 38, pp. 13310–13319, 2011. G. K. H. Pang and K. Takahashi, “Adaptive Route Selection for Dynamic Route Guidance System based on Fuzzy-Neural Approaches,” IEEE Transactions on Vehicular Technology, vol. 48, p. 2028, 1999. L. Liu, J. Xu, S. S. Liao, and H. Chen, “A real-time personalized route recommendation system for self-drive tourists based on vehicle to vehicle communication,” Expert Systems with Applications, vol. 41, no. 7, pp. 3409–3417, Jun. 2014. R. A. Abbaspour and F. Samadzadegan, “Timedependent personal tour planning and scheduling in metropolises,” Expert Systems with Applications, vol. 38, no. 10, pp. 12439–12452, Sep. 2011. L. Castillo, E. Armengol, E. Onaindia, L. Sebastia, J. Gonzalez-Boticario, A. Rodriguez, S. Fernandez, J. D. Arias, and D. Borrajo, “SAMAP: An useroriented adaptive system for planning tourist visits,” EXPERT SYSTEMS WITH APPLICATIONS, vol. 34, pp. 1318–1332, 2008. I. Horrocks, “Ontologies and the Semantic Web,” Commun. ACM, vol. 51, no. 12, pp. 58–67, Dec. 2008. B. Rodríguez, J. Molina, F. Pérez, and R. Caballero, “Interactive design of personalised tourism routes,” Tourism Management, vol. 33, no. 4, pp. 926–940, Aug. 2012. B. Petrevska and S. Koceski, “Tourism Recommendation System: Empirical Investigation,” Revista de Turism - Studii si Cercetari in Turism, no. 14, pp. 11–18, Dec. 2012. C.-S. Lee, Y.-C. Chang, and M.-H. Wang, “Ontological recommendation multi-agent for Tainan City travel,” Expert Systems with Applications, vol. 36, no. 3, Part 2, pp. 6740–6753, Apr. 2009. P.-O. Siebers and U. Aickelin, “Introduction to Multi-Agent Simulation,” Mar. 2008. S. Payr, P. Petta, and R. Trappl, Emotions in Humans and Artifacts. Cambridge, Mass: MIT Press, 2002. R. Chu, “What online Hong Kong travelers look for on airline/travel websites?,” International Journal of Hospitality Management, vol. 20, no. 1, pp. 95–100, 01 2001. H.-S. Chiang and T.-C. Huang, “User-adapted travel planning system for personalized schedule recommendation,” Information Fusion. A. Montejo-Ráez, J. M. Perea-Ortega, M. Á. García-Cumbreras, and F. Martínez-Santiago,
[38]
[39]
[40]
[41]
[42]
[43] [44]
[45]
[46]
[47]
[48] [49]
“Otiŭm: A web based planner for tourism and leisure,” Expert Systems with Applications, vol. 38, pp. 10085–10093, 2011. A. Moreno, A. Valls, D. Isern, L. Marin, and J. Borràs, “SigTur/E-Destination: Ontology-based personalized recommendation of Tourism and Leisure Activities,” Engineering Applications of Artificial Intelligence, vol. 26, pp. 633–651, 01 / 01 /. D. Buhalis and R. Law, “Progress in information technology and tourism management: 20 years on and 10 years after the Internet—The state of eTourism research,” Tourism Management, vol. 29, no. 4, pp. 609–623, Aug. 2008. W.-S. Yang and S.-Y. Hwang, “iTravel: A recommender system in mobile peer-to-peer environment,” Journal of Systems and Software, vol. 86, no. 1, pp. 12–20, Jan. 2013. K. ten Hagen, R. Kramer, M. Hermkes, B. Schumann, and P. Mueller, “Semantic Matching and Heuristic Search for a Dynamic Tour Guide,” in Information and Communication Technologies in Tourism 2005, D. A. J. Frew, Ed. Springer Vienna, 2005, pp. 149–159. D. Gavalas, C. Konstantopoulos, K. Mastakas, and G. Pantziou, “A survey on algorithmic approaches for solving tourist trip design problems,” J Heuristics, pp. 1–38. B. L. Golden, L. Levy, and R. Vohra, “The orienteering problem,” Naval Research Logistics, vol. 34, no. 3, p. 307, Jun. 1987. P. ( 1 ) Vansteenwegen, W. ( 1 Souffriau 2 ), D. ( 1 ) Van Oudheusden, and G. ( 2 ) Vanden Berghe, “Iterated local search for the team orienteering problem with time windows,” Computers and Operations Research, vol. 36, no. 12, pp. 3281– 3290, 01 2009. P. Vansteenwegen, W. Souffriau, G. V. Berghe, and D. V. Oudheusden, “The City Trip Planner: An expert system for tourists,” Expert Systems with Applications, vol. 38, no. 6, pp. 6540–6546, Jun. 2011. D. Gavalas, C. Konstantopoulos, K. Mastakas, and G. Pantziou, “Mobile recommender systems in tourism,” Journal of Network and Computer Applications. W. Souffriau and P. Vansteenwegen, “Tourist Trip Planning Functionalities: State–of–the–Art and Future,” in Current Trends in Web Engineering, F. Daniel and F. M. Facca, Eds. Springer Berlin Heidelberg, 2010, pp. 474–485. R. Cattell, “Scalable SQL and NoSQL Data Stores,” SIGMOD Rec., vol. 39, no. 4, pp. 12–27, May 2011. I. Garcia, L. Sebastia, and E. Onaindia, “On the design of individual and group recommender systems for tourism,” Expert Systems with Applications, vol. 38, no. 6, pp. 7683–7692, Jun. 2011.
