Enhancing User Experience in Smart Tourism via Fuzzy Logic-Based Personalization

Abstract

In the era of smart tourism, providing seamless and personalized experiences has become significant for enhancing user satisfaction and engagement. This paper presents a novel fuzzy logic-based application system designed to enhance personalization in smart tourism. The proposed system integrates real-time user data and delivers customized services to each particular user. In particular, the proposed system incorporates a recommendation mechanism that combines TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) with fuzzy logic to assess multiple criteria and user preferences and provide accurate and well-rounded personalized travel destination recommendations. By employing fuzzy logic, the system effectively overcomes challenges associated with uncertainty and subjectivity in user data, enabling precise and adaptable decision-making and ensuring more accurate service recommendations. Through case studies and simulations, the paper evaluates the system’s impact on enhancing user satisfaction and the overall tourism experience. Furthermore, preliminary evaluation results demonstrate the system’s ability to generate meaningful and seamless personalized recommendations that enhance the provided tourism services. This work contributes to the growing field of smart tourism by offering a scalable and user-centric solution. The scalability of the system is ensured through its efficient handling of multidimensional data, adaptability to diverse user profiles, and extendability to various tourism applications, including destination ranking, activity recommendations, and hotel selection. Additionally, its integration potential with existing travel platforms highlights its applicability in real-world scenarios, making it a robust tool for enhancing smart-tourism experiences.

1. Introduction

Smart Tourism is a term that has been around for a while, with its first appearance being dated in the late 1990s [1]. In the related literature, though, we locate a growing interest in the topic about a decade ago [2,3,4]. Since then, interest in the area has steadily expanded, as demonstrated by the hundreds of academic results on Google Scholar and a rising number of publications in Scopus [5]. Actually, Google Scholar lists over 39,000 results for “Smart Tourism”, with an increasing trend in annual publications.

At the same time, there is no denying that in this day and age, when personalization appears to occur everywhere, the desire for travel experiences that are smooth, user-centric, and personalized has highly increased. With such an information overload, travelers usually face difficulties due to various options available that do not always satisfy their own diverse preferences. That being the case, numerous researchers and practitioners seek to address these issues; however, many areas for research are still available.

One such approach that could demonstrate exceptional promise in tackling the aforementioned challenges is fuzzy logic. Unlike traditional AI methods, fuzzy logic mimics human reasoning by “calculating with words”, allowing it to handle ambiguity and imprecision effectively [6,7]. Moreover, fuzzy logic is distinguished by its strengths in explainability, interpretability, and transparency, making its decision-making processes both accessible and comprehensible to users [8,9]. These unique abilities of fuzzy logic make it significant in the development of effective smart applications and systems. Therefore, it can be used in smart tourism to model the diverse user needs and preferences to offer a level of personalization that enhances the overall tourism experience. In addition, personalized recommendations in smart tourism are based on the assessment and comparison of multiple criteria, like costs, travel time, safety, quality of life, opportunities for activities, heritage culture, etc. This is succeeded using a multiple-criteria decision-making technique, like TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) [10]. Regarding the above, the combination of TOPSIS with fuzzy logic is especially effective for managing the complex and ambiguous process of generating personalized recommendations in smart tourism [11,12,13]. In particular, fuzzy TOPSIS could provide robust and scalable solutions for enhancing smart-tourism systems by combining and modeling diverse, vague, and subjective data inputs.

While previous studies have explored fuzzy logic and multiple-criteria decision-making (MCDM) techniques in tourism, most approaches either focus on static preference modeling or lack real-time adaptability. Our approach builds upon fuzzy TOPSIS but introduces autonomous user adaptation, ensuring that recommendations evolve dynamically based on real-time data. Moreover, our system is designed for seamless integration with travel industry platforms, making it scalable and practically deployable. This combination of flexibility, adaptability, and integration potential distinguishes our approach from traditional tourism recommendation systems, which often rely on rule-based or purely data-driven methods that lack interpretability and personalization.

Based on the above, this paper introduces a fuzzy logic-based personalization system, which is based on fuzzy TOPSIS, designed to optimize user experiences in smart tourism. The system leverages a comprehensive dataset that includes key indicators, namely Cost of Living, Quality of Life, Adventure, Heritage, Average Travel Cost, Travel Time, and Safety. Based on these criteria 46 and user-specific preferences, the system generates a ranked list of countries, providing personalized recommendations tailored to the unique needs of each traveler.

In addition to offering targeted suggestions, the system has the potential to operate autonomously, allowing it to adapt dynamically to real-time data and change user inputs without requiring constant oversight. Furthermore, the system can be seamlessly integrated into existing platforms such as travel agency software, booking systems, and tourism recommendation applications, enhancing their capabilities with robust and scalable personalized recommendations. By combining multidimensional data with a fuzzy TOPSIS-based framework, the proposed system delivers a versatile and innovative solution that addresses the complexities of decision-making in the tourism sector. Through detailed case studies and simulations, we evaluate the system’s ability to deliver meaningful and seamless travel experiences, contributing to the ongoing evolution of smart-tourism technologies.

The remainder of the paper is organized as follows: Section 2 provides an overview of previous research in smart-tourism personalization, recommender systems, and the application of fuzzy logic in decision-making, identifying gaps and highlighting the need for the proposed approach. Section 3 describes the architecture of the fuzzy logic-based personalization system, detailing the key criteria used, the dataset structure, and the algorithm for ranking travel destinations based on user preferences. Section 4 illustrates real-world applications of the system, demonstrating how it generates personalized travel recommendations and adapts to dynamic user inputs. Section 5 assesses the system’s performance through experiments and user studies, measuring its accuracy, effectiveness, and user satisfaction. Section 6 critically analyzes the findings, comparing the system with traditional recommendation methods and discussing potential improvements. Finally, Section 7 summarizes the study’s contributions, emphasizing the advantages of the proposed system and outlining future research directions, such as integrating the system into existing travel platforms for broader adoption.

2. Related Work

2.1. Literature Review

2.1.1. Smart Tourism, Personalization, and Recommender Systems

Smart tourism is commonly conceptualized as the use of advanced technologies to create a digital ecosystem aimed at enhancing and optimizing tourism for the benefit of industries, travelers, local communities, nations, and destinations. By leveraging infrastructure such as sensors, cloud computing, smartphones, RFID, and Wi-Fi, both researchers and industry practitioners are exploring various methodologies within the realm of smart tourism [14]. Simultaneously, innovations in algorithms and emerging technologies are being researched and their capability to help deliver smart-tourism services. These include recommender systems that suggest attractions tailored to a user’s profile [15], systems that analyze traveler locations to infer behavior and enable location-based advertising [14], and applications utilizing sensor data to enhance tourism experiences [16,17]. However, while these technological advancements provide vast amounts of data, the challenge remains in effectively processing and interpreting them to offer meaningful recommendations.

A key advancement in smart tourism has been the integration of AI-powered recommender systems, which tailor travel experiences to user preferences. Previous research has explored attraction-based recommendations [15], location-based analytics for behavior inference [14], and sensor-driven tourism enhancement [16,17]. The growing adoption of AI-driven personalization has also led to the use of automated travel assistants, customer service bots, and predictive analytics [18,19,20]. Further improving smart-tourism applications [21,22] are new developments in Large Language Models (LLMs) such as GPTs, showing potential for real-time travel suggestions and language translation.

Despite these advancements, significant challenges remain in personalizing travel experiences. Travelers frequently face a flood of information, a lack of expertise to make informed choices, and circumstances where the timing or location is not ideal for the best decisions to thrive [15,23,24,25]. Recommender systems are ideally suited to tackle these challenges by providing personalized suggestions that streamline and improve the decision-making process [26,27]. While existing recommender systems effectively utilize heterogeneous data sources, such as service attributes [28], contextual factors like time, weather, and user budget [27], and cultural heritage databases [16,29,30,31,32], they often rely on static models that lack adaptability to real-time user behavior and evolving preferences. Additionally, traditional recommender systems struggle to handle subjectivity and uncertainty in user preferences, which can significantly impact the relevance of recommendations.

Despite the growing advancements in smart tourism, travelers still face a fundamental challenge: how to make the best choice in an overwhelming sea of options. While recommendation systems have helped streamline decision-making, they often rely on fixed rules or rigid data models that struggle to account for the fluidity and subjectivity of human preferences. A budget-conscious traveler may prioritize cost, while another values time efficiency, and yet another seeks the safest possible destination. Existing solutions tend to oversimplify these trade-offs, often missing the nuances that shape real-world travel decisions. Our research seeks to bridge this gap by introducing a fuzzy TOPSIS-based recommendation system designed to reflect the way people make travel decisions—by balancing multiple, sometimes conflicting, factors in an intuitive way.

2.1.2. Fuzzy Logic in User-Centric Systems

To tackle the uncertainty and complexity in travel decision-making, fuzzy logic has been proposed as a promising approach to smart tourism. Its capacity to handle unclear or ambiguous information makes it especially effective at creating tailored travel suggestions and improving user experiences. Several studies have demonstrated its effectiveness in improving personalized travel experiences. For example, research has explored how fuzzy logic can be integrated with blockchain-based cryptocurrency systems to provide secure and transparent travel transactions [33].

Moreover, another related research is illustrated in Mei et al. [34], where the authors introduce the design and execution of an Intelligent Tourism Management System (ITMS) utilizing Fuzzy Cluster Analysis (FCA). FCA is designed to organize and analyze complex tourist data using fuzzy logic and clustering methods. The goal of this approach is to allow systems to adapt to the uncertainties in tourist preferences and behaviors, offering customized experiences that cater to each person’s unique needs. Further applications of fuzzy logic in tourism are illustrated in [35], where a fuzzy control system is employed to assist tourists in selecting accommodations and travel destinations. The algorithm compares hotels in detail and assesses service quality based on sentiment analysis and criteria, including location ratings, service, and food quality.

Another significant contribution is presented in [36], where data sparsity is addressed in travel recommendation agents, a challenge exacerbated by the COVID-19 pandemic. Their research integrates fuzzy logic with clustering and dimensionality reduction techniques to improve recommendation accuracy. By evaluating their neuro-fuzzy approach to TripAdvisor data, they demonstrate how fuzzy logic can effectively mitigate sparsity issues, enhancing the reliability and personalization of travel recommendations. While these studies highlight the potential of fuzzy logic in tourism, they do not focus on real-time personalized recommendations based on evolving user needs.

That being the case, the transformative potential of fuzzy logic in smart tourism is quite evident. From generating accurate recommendations to overcoming technical challenges like data sparsity, fuzzy logic offers a robust framework for creating intelligent and user-centric tourism systems. However, the provision of personalized travel destination recommendations requires the simultaneous assessment and comparison of multiple criteria, like costs, travel time, safety, quality of life, opportunities of activities, cultural heritage, etc. This is managed by multiple-criteria decision-making techniques, which provide a structured decision-making process that systematically evaluates and prioritizes multiple, often conflicting criteria to arrive at the best possible choice or solution [37]. Nevertheless, user preferences and data concerning tourism choices are often subjective and imprecise. Fuzzy logic is ideal to deal with this issue. Therefore, a solution for more accurate and well-rounded personalized travel destination recommendations is to combine fuzzy logic with multiple-criteria decision-making technique [38]. A such method is fuzzy TOPSIS (Technique for Order Preference by Similarity to Ideal Solution), which is a powerful decision-making method that combines the strengths of fuzzy logic and the TOPSIS approach, making it particularly well-suited for handling complex, subjective, and uncertain data in multicriteria decision-making problems, like personalized recommendations in smart tourism. Regarding the above, this research introduces a fuzzy TOPSIS-based recommendation system that captures the uncertainty and subjectivity of user preferences, incorporates multiple conflicting criteria such as cost, safety, and travel time in a structured decision-making approach, dynamically adapts recommendations based on real-time user inputs, and ensures seamless integration with existing tourism platforms. By bridging the gap between static AI-driven recommender systems and multicriteria fuzzy logic decision-making, this work advances personalized smart-tourism recommendations, offering a solution that is both theoretically robust and practically applicable.

2.2. Fuzzy TOPSIS

Fuzzy TOPSIS is a fuzzy logic-based multicriteria decision-making method. It is an extension of the classical TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) method, which was introduced by Ching-Lai Hwang and Yoon in 1981 [39]. It incorporates fuzzy logic to handle uncertainty and imprecision in decision-making processes. It is widely used in decision-making situations where challenges in quantifying preferences or judgments precisely are faced. By incorporating fuzzy logic, it allows decision-makers to represent subjective preferences and ambiguous information using linguistic terms and fuzzy numbers, providing clear and interpretable rankings of alternatives. The steps of fuzzy TOPSIS follow:

• Step 1: Identify alternatives and criteria.
• Step 2: Construct a Fuzzy Decision Matrix, which represents criteria evaluations using fuzzy numbers for each alternative.
• Step 3: Normalize the Fuzzy Decision Matrix.
• Step 4: Construct the Weighted Normalized Fuzzy Decision Matrix by applying weights to criteria based on their importance.
• Step 5: Calculate the Fuzzy Positive Ideal Solution (FPIS) and the Fuzzy Negative Ideal Solution (FNIS).
• Step 6: Compute the distances of each alternative from the FPIS and FNIS.
• Step 7: Compute the closeness coefficient for each alternative to determine the ranking.
• Step 8: Rank the alternatives and select the alternative with the highest closeness coefficient, which represents the best alternative.

3. System Design

3.1. Data

For our research, we created and utilized a dataset that provides valuable insights into the realm of smart tourism. This dataset, which is presented in

Table 1. Smart-Tourism Dataset.

Country	Cost-of-Living Index	Adjusted Quality of Life Ranking	Adjusted Adventure Ranking	Adjusted Heritage Ranking	Average Travel Cost (EUR)	Travel Time (h)	Safety Index
Denmark	72.3	2	12	12	380	3.42	73.7
Austria	65.1	8	8	7	135	2.17	70.7
Ireland	64.4	10	6	6	292.5	6.67	53.5
France	63.7	12	4	3	255	3.58	44.7
Finland	63.2	3	11	19	522.5	3.83	73.7
The Netherlands	63.1	5	7	10	287.5	3.5	73.7
Luxembourg	62.4	11	18	21	427.5	3.08	65.6
Germany	62.2	6	17	8	337.5	3.0	61.1
United Kingdom	62.0	9	15	5	150	3.92	52.6
Belgium	61.1	7	14	15	190	3.42	50.8
Sweden	59.3	1	9	16	252.5	5.5	51.6
Italy	56.2	15	1	1	132.5	2.17	52.7
Cyprus	54.7	24	10	18	107.5	1.58	67.6
Estonia	52.0	21	23	24	300	7.0	76.2
Slovenia	49.9	23	22	23	257.5	6.0	75.8
Latvia	49.1	26	25	25	297.5	6.5	62.6
Spain	47.3	13	2	2	195	3.92	63.8
Lithuania	47.1	25	26	26	240	5.0	67.2
Slovakia	46.7	22	24	22	147.5	2.25	68.9
Croatia	45.5	17	13	14	225	5.92	74.0
Portugal	45.1	14	3	4	400	4.5	68.8
Hungary	41.7	18	20	13	107.5	2.17	66.3
Poland	40.8	16	16	11	165	2.67	69.3
Bulgaria	38.3	20	21	20	140	1.5	62.9
Romania	37.5	19	19	17	132.5	1.5	66.3
Switzerland	101.1	4	5	9	220	2.8	73.5

, encompasses a range of indicators, including the Cost of Living Index Adjusted Quality of Life Ranking, Adjusted Adventure Ranking, Adjusted Heritage Ranking, Average Price, Travel Time, and Safety Index. These indicators are crucial for understanding and evaluating the economic, social, cultural, and safety aspects of countries, making them highly relevant in the context of smart tourism.

Unlike rankings linked to entrepreneurship or power, which mostly represent economic influence and corporate dynamics, the chosen indicators concentrate on elements that directly affect the travel experience and decision-making of visitors. Smart tourism emphasizes affordability, accessibility, cultural richness, safety, and overall visitor satisfaction. By including these aspects, our dataset offers a more complete view of how locations meet visitors’ demands in an environment of a digitalized, experience-driven smart-tourism landscape.

In this section, we elaborate on the origins of these data, detailing how each indicator was obtained and calculated. Additionally, we explain the methodologies used to analyze and adapt the data to suit the scope of our research, ensuring that the findings align with the objectives of understanding tourism dynamics in a European context.

Initially, data for flights from Athens to the capitals of European countries were requested from ChatGPT to obtain a sample for analysis. These data included the average travel time duration and the average price. However, recognizing the limitations of AI-generated estimations, we conducted a thorough comparison with actual flight data from the Google Travel website (Google. (2025). Google Travel. Retrieved from https://www.google.com/travel/explore?tfs=CBwQAxobEgoyMDI1LTAxLTI0KABqCwgCEgcvbS8wb (accessed on 27 February 2025)), where significant differences were identified. This discrepancy was primarily due to ChatGPT’s estimation of distances between locations, which does not necessarily translate into accurate travel time. For many destinations, long layovers or the absence of direct routes influenced the actual travel duration. To address this issue, the travel times were recalculated based on available flights to the destinations using additional research on platforms like Skyscanner (Skyscanner. (2025). Skyscanner Flights. Retrieved from https://www.skyscanner.net (accessed on 27 February 2025)), where destinations with standard layovers were also reviewed. Once again, similar results were observed, indicating that relying solely on distance-based calculations often underestimates the total travel time.

This distinction is critical in the context of travel planning and analysis, as realistic travel time estimates provide more accurate insights for decision-making in tourism. By recalculating the average travel time and presenting it in the column Travel Time (h) in the table, we ensured that the data reflects actual conditions, which is essential for understanding the accessibility and convenience of different destinations in the smart-tourism domain.

From the Google Travel website, we were also able to extract the typical price range for each destination. For example, in the case of flights to Stockholm, the website provided a section that highlighted that the least expensive flights for trips to Stockholm typically cost between EUR 170–305. By using this information, we were able to calculate the average price for each destination, which is presented in the Average Price (EUR) column of the table. This calculation allows for a more standardized comparison of flight costs across different destinations, providing valuable insights into the affordability of travel in the context of smart tourism.

The Safety Index provides a numerical representation of the relative safety of a country, with higher values indicating greater levels of safety. This metric is derived from various factors such as crime rates, public perception of safety, and governmental stability, and is widely utilized in comparative studies of global safety (Data for the Safety Index were sourced from Numbeo, a comprehensive database that aggregates user-submitted information on safety and crime rates across various countries. For more details, see (https://www.numbeo.com/crime/rankings_by_country.jsp accessed on 27 February 2025). The index ranges from 0 to 100, with countries such as Estonia and Slovenia scoring high for their well-maintained law enforcement systems, whereas others, like France and the United Kingdom, present lower scores due to specific challenges related to urban safety. As such, the Safety Index is instrumental in understanding socio-political environments and assessing risks associated with international travel and residence.

An important index of the utilized dataset is the Cost-of-Living Index that was extracted from the “Cost-of-Living Index by Country 2024” dataset that can be found on Kaggle (accessed on 27 February 2025) (https://www.kaggle.com/datasets/myrios/cost-of-living-index-by-country-by-number-2024?resource=downloadCost_of_Living_Index_by_Country_2024.csv). The original data were sourced from Numbeo, (accessed on 27 February 2025) (https://www.numbeo.com/cost-of-living/cpi_explained.jsp) which provides the relative cost-of-living indicators in comparison to New York City, which is set as the baseline index of 100%.

Excluding housing costs like rent or mortgage, the relative prices of consumer goods like groceries, restaurants, transportation, and utilities are measured by the Cost-of-Living Index. To illustrate the point, a city that has a 120 on the Cost-of-Living Index is predicted to be 20% more costly than New York City (rent excluded). The cost of living in various countries can be normalized using this index. Because of its relative character and the fact that it does not include the cost of accommodation, it is excellent for comparing regional economic situations and the cost of basic necessities. As part of our research, this index sheds light on the European economic landscape, which is useful for understanding how much money tourists can expect to spend when visiting different countries—information that can be used to make sensible tourism planning decisions.

Part of the exploited data is ranking, specifically the Quality of Life Ranking, the Adventure Ranking, and the Heritage Ranking. These data were obtained from the Best Countries 2023 Overall Rankings report released by News (accessed on 27 February 2025) (https://www.usnews.com/media/best-countries/2023-overall-rankings.pdf). The rankings, created in partnership with the Wharton School at the University of Pennsylvania and BAV Group, assess countries by analyzing global perceptions across 73 attributes, which are categorized into 10 thematic subrankings, including Quality of Life, Adventure, and Heritage. The scoring of these attributes was derived from survey responses collected from more than 17,000 individuals across the globe, encompassing informed elites, business decision-makers, and the general public, which guarantees a diverse and representative sample. In comparison to the other information utilized in our system, the lower the ranking of a country, the higher it scores. To exemplify, Italy has the top Heritage ranking. Thus, it is the best one when it comes to heritage influence from the others included in our dataset. The extensive nature of the sample and the thoroughness of the methodology ensure that the resulting rankings offer a dependable and strong assessment of a country’s performance across different qualitative aspects. Our confidence in the rankings enables us to use them as a solid basis for additional analysis in this study.

The survey utilized in this study features numerous rankings, including Agility, Entrepreneurship, Quality of Life, Movers, Social Purpose, Cultural Influence, Open for Business, Power, Adventure, and Heritage. From these, we selected the Quality of Life, Adventure, and Heritage rankings, as they align most closely with the scope of our research, particularly in understanding aspects of smart tourism and cultural engagement.

To be more precise, the Quality of Life Ranking evaluates attributes such as the availability of a robust job market, economic stability, affordability, political stability, safety, and access to well-developed education and healthcare systems. The Adventure Ranking highlights qualities like a country’s friendliness, suitability for tourism, pleasant climate, scenic landscapes, and general appeal. Meanwhile, the Heritage Ranking emphasizes cultural richness, historical significance, culinary traditions, and the presence of both cultural and geographical attractions. In the context of smart tourism and regional analysis, these subrankings collectively shed light on the many appealing aspects of each country.

We concentrated our research on a specific group of 25 European countries chosen from a broader dataset that included 87 nations. To refine the analysis for this more specific focus, the Quality of Life Ranking, Adventure Ranking, and Heritage Ranking were updated to better represent the relative positions of these countries within our dataset. This adjustment guarantees that the results are pertinent to the European setting, facilitating a more profound comprehension of the distinct regional dynamics involved. The revised rankings provide insights into how perceptions shape tourism trends and cultural interactions in Europe.

3.2. The Fuzzy Logic-Based Mechanism for Personalized Recommendations

Fuzzy TOPSIS can handle both subjective and objective criteria effectively, making it well-suited for modeling user preferences in a personalized tourism system. Unlike other multiple-criteria decision-making techniques, like AHP, Fuzzy TOPSIS allows for a more flexible evaluation process by directly ranking alternatives based on their closeness to an ideal solution [40,41]. Additionally, while Machine Learning models can be powerful for personalization, they often require extensive labeled data and may lack interpretability, whereas Fuzzy TOPSIS provides a structured yet adaptable decision-making framework that accommodates uncertainty in user preferences [42,43]. Given the above, the proposed mechanism for personalized recommendations is based on Fuzzy TOPSIS.s

3.2.1. Fuzzy Sets for Criteria

Fuzzy logic aims to deal with uncertainty, imprecision, and subjectivity that characterize the processes of reasoning and decision-making in real-world scenarios. For this reason, fuzzy logic does not use crisp values, but it uses linguistic values that imitate the human way of thinking. Therefore, the decision criteria of the proposed system must be defined through fuzzy sets. The defined fuzzy sets for the variables of the presented system are described in

Table 2. Fuzzy sets for criteria.

Criterion	Linguistic Term	Membership Partition	Function
Cost-of-Living Index	Very Low (VL)	(0, 0, 20, 30)
	Low (L)	(20, 30, 40, 50)
	Moderate (M)	(40, 50, 60, 70)
	High (H)	(60, 70, 85, 100)
	Very High (VH)	(85, 100, 120, 120)
Adjusted Quality of Life Ranking	Very High (VH)	(1, 1, 3, 5)
	High (H)	(3, 5, 8, 10)
	Moderate (M)	(8, 10, 14, 17)
	Low (L)	(14, 17, 20, 23)
	Very Low (VL)	(20, 23, 26, 26)
Adjusted Adventure Ranking	Very High (VH)	(1, 1, 3, 5)
	High (H)	(3, 5, 8, 10)
	Moderate (M)	(8, 10, 14, 17)
	Low (L)	(14, 17, 20, 23)
	Very Low (VL)	(20, 23, 26, 26)
Adjusted Heritage Ranking	Very High (VH)	(1, 1, 3, 5)
	High (H)	(3, 5, 8, 10)
	Moderate (M)	(8, 10, 14, 17)
	Low (L)	(14, 17, 20, 23)
	Very Low (VL)	(20, 23, 26, 26)
Average Travel Cost	Very Low (VL)	(0, 0, 50, 100)
	Low (L)	(50, 100, 130, 170)
	Moderate (M)	(150, 200, 250, 300)
	High (H)	(280, 300, 350, 380)
	Very High (VH)	(350, 450, 600, 600)
Travel Time	Very Low (VL)	(0, 0, 1, 1.3)
	Low (L)	(1, 1.3, 1.5, 2)
	Moderate (M)	(1.5, 2, 3, 3.5)
	High (H)	(3, 3.5, 4, 5)
	Very High (VH)	(4, 5, 6, 6)
Safety Index	Very Low (VL)	(0, 0, 10, 20)
	Low (L)	(10, 20, 30, 40)
	Moderate (M)	(30, 40, 60, 70)
	High (H)	(60, 70, 80, 90)
	Very High (VH)	(80, 90, 100, 100)

and Figure A1–Figure A7 in Appendix A.

3.2.2. Linguistic Terms for Criteria Ratings and Weights

In Fuzzy TOPSIS each criterion is rated with a linguistic term. Furthermore, the linguistic terms and their scaling of the ratings must be similar for all criteria for homogeneity and easy computation reasons. Furthermore, fuzzy TOPSIS allows the definition of weights of significance for each criterion. These weights are expressed, also, as linguistic terms. In

Table 3. Fuzzy sets for criteria ratings of alternatives.

Linguistic Term	Partition of the Triangular Membership Function
Very Low (VL)	(1, 1, 3)
Low (L)	(1, 3, 5)
Moderate (M)	(3, 5, 7)
High (H)	(5, 7, 10)
Very High (VH)	(7, 10, 10)

and

Table 4. Fuzzy sets for weights.

Linguistic Term	Partition of the Triangular Membership Function
None	(0, 0, 0)
Very Low (VL)	(1, 1, 30)
Low (L)	(1, 30, 50)
Medium (M)	(30, 50, 70)
High (H)	(50, 70, 100)
Very High (VH)	(70, 100, 100)

, the fuzzy sets for criteria ratings and weights are presented accordingly.

3.3. Fuzzy TOPSIS-Based Mechanism

Step 1: Convert crisp values of criteria to fuzzy values.

Table 5. Linguistic values for each alternative.

Country	Cost-of-Living Index	Adjusted Quality of Life Ranking	Adjusted Adventure Ranking	Adjusted Heritage Ranking	Average Travel Cost (EUR)	Travel Time	Safety Index
Denmark	H	VH	M	M	VH	M & H	H
Austria	M & H	H	H	H	L	M	H
Ireland	M & H	M	H	H	M & H	VH	M
France	M & H	M	H & VH	VH	M	H	M
Finland	M & H	VH	M	L	VH	H	H
The Netherlands	M & H	H	H	M	H	H	H
Luxembourg	M & H	M	L	VL & L	VH	M & H	M & H
Germany	M & H	H	L	H	H	M	M & H
United Kingdom	M & H	M & H	L & M	H	L	H	M
Belgium	M & H	H	M	L & M	M	M & H	M
Sweden	M	VH	M & H	M	M	VH	M
Italy	M	L & M	VH	VH	L	M	M
Cyprus	M	VL	L	L	L	L & M	M & H
Estonia	M	VL & L	VL	VL	H	VH	H
Slovenia	L & M	VL	VL	VL	M	VH	H
Latvia	L & M	VL	VL	VL	M & H	VH	M & H
Spain	L & M	M	VH	VH	M	H	M & H
Lithuania	L & M	VL	VL	VL	M	VH	M & H
Slovakia	L & M	VL & L	VL & L	VL & L	L	M	M & H
Croatia	L & M	L	M	M	M	VH	H
Portugal	L & M	M	H & VH	H & VH	VH	H & VH	M & H
Hungary	L & M	L	M	M	L	M	M & H
Poland	L & M	M	M	M	L and M	M	M & H
Bulgaria	L	L	L	L	L	L	M & H
Romania	L	L	L	L	L	L	M & H
Switzerland	VH	H & VH	M & H	M and H	M	M	H

presents the fuzzy values of the criteria from the countries of the dataset, which is used for our research. They were calculated taking into account the above defined fuzzy sets.

Step 2: Replace the linguistic terms of the cells of Table 5 with a set of three numbers (a, b, c), which represents the partition of each fuzzy set. This is accomplished using the fuzzy sets of ratings (Table 3). The produced matrix D is called a fuzzy decision matrix. The result of this step is

Table A1. Alternatives Ratings with Fuzzy Numbers.

Country	Cost-of-Living Index	Adjusted Quality of Life Ranking	Adjusted Adventure Ranking	Adjusted Heritage Ranking	Average Travel Cost (EUR)	Travel Time	Safety Index
Denmark	(5, 7, 10)	(7, 10, 10)	(3, 5, 7)	(3, 5, 7)	(7, 10, 10)	(3, 5, 7) & (5, 7, 10)	(5, 7, 10)
Austria	(3, 5, 7) & (5, 7, 10)	(5, 7, 10)	(5, 7, 10)	(5, 7, 10)	(1, 3, 5)	(3, 5, 7)	(5, 7, 10)
Ireland	(3, 5, 7) & (5, 7, 10)	(3, 5, 7)	(5, 7, 10)	(5, 7, 10)	(3, 5, 7) & (5, 7, 10)	(7, 10, 10)	(3, 5, 7)
France	(3, 5, 7) & (5, 7, 10)	(3, 5, 7)	(5, 7, 10) & (7, 10, 10)	(7, 10, 10)	(3, 5, 7)	(5, 7, 10)	(3, 5, 7)
Finland	(3, 5, 7) & (5, 7, 10)	(7, 10, 10)	(3, 5, 7)	(1, 3, 5)	(7, 10, 10)	(5, 7, 10)	(5, 7, 10)
The Netherlands	(3, 5, 7) & (5, 7, 10)	(5, 7, 10)	(5, 7, 10)	(3, 5, 7)	(5, 7, 10)	(5, 7, 10)	(5, 7, 10)
Luxembourg	(3, 5, 7) & (5, 7, 10)	(3, 5, 7)	(1, 3, 5)	(1, 1, 3) & (1, 3, 5)	(7, 10, 10)	(3, 5, 7) & (5, 7, 10)	(3, 5, 7) & (5, 7, 10)
Germany	(3, 5, 7) & (5, 7, 10)	(5, 7, 10)	(1, 3, 5)	(5, 7, 10)	(5, 7, 10)	(3, 5, 7)	(3, 5, 7) & (5, 7, 10)
United Kingdom	(3, 5, 7) & (5, 7, 10)	(3, 5, 7) & (5, 7, 10)	(1, 3, 5) & (3, 5, 7)	(5, 7, 10)	(1, 3, 5)	(5, 7, 10)	(3, 5, 7)
Belgium	(3, 5, 7) & (5, 7, 10)	(5, 7, 10)	(3, 5, 7)	(1, 3, 5) & (3, 5, 7)	(3, 5, 7)	(3, 5, 7) & (5, 7, 10)	(3, 5, 7)
Sweden	(3, 5, 7)	(7, 10, 10)	(3, 5, 7) & (5, 7, 10)	(3, 5, 7)	(3, 5, 7)	(7, 10, 10)	(3, 5, 7)
Italy	(3, 5, 7)	(1, 3, 5) & (3, 5, 7)	(7, 10, 10)	(7, 10, 10)	(1, 3, 5)	(3, 5, 7)	(3, 5, 7)
Cyprus	(3, 5, 7)	(1, 1, 3)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5) & (3, 5, 7)	(3, 5, 7) & (5, 7, 10)
Estonia	(3, 5, 7)	(1, 1, 3) & (1, 3, 5)	(1, 1, 3)	(1, 1, 3)	(5, 7, 10)	(7, 10, 10)	(5, 7, 10)
Slovenia	(1, 3, 5) & (3, 5, 7)	(1, 1, 3)	(1, 1, 3)	(1, 1, 3)	(3, 5, 7)	(7, 10, 10)	(5, 7, 10)
Latvia	(1, 3, 5) & (3, 5, 7)	(1, 1, 3)	(1, 1, 3)	(1, 1, 3)	(3, 5, 7) & (5, 7, 10)	(7, 10, 10)	(3, 5, 7) & (5, 7, 10)
Spain	(1, 3, 5) & (3, 5, 7)	(3, 5, 7)	(7, 10, 10)	(7, 10, 10)	(3, 5, 7)	(5, 7, 10)	(3, 5, 7) & (5, 7, 10)
Lithuania	(1, 3, 5) & (3, 5, 7)	(1, 1, 3)	(1, 1, 3)	(1, 1, 3)	(3, 5, 7)	(7, 10, 10)	(3, 5, 7) & (5, 7, 10)
Slovakia	(1, 3, 5) & (3, 5, 7)	(1, 1, 3) & (1, 3, 5)	(1, 1, 3) & (1, 3, 5)	(1, 1, 3) & (1, 3, 5)	(1, 3, 5)	(3, 5, 7)	(3, 5, 7) & (5, 7, 10)
Croatia	(1, 3, 5) & (3, 5, 7)	(1, 3, 5)	(3, 5, 7)	(3, 5, 7)	(3, 5, 7)	(7, 10, 10)	(5, 7, 10)
Portugal	(1, 3, 5) & (3, 5, 7)	(3, 5, 7)	(5, 7, 10) & (7, 10, 10)	(5, 7, 10) & (7, 10, 10)	(7, 10, 10)	(5, 7, 10) & (7, 10, 10)	(3, 5, 7) & (5, 7, 10)
Hungary	(1, 3, 5) & (3, 5, 7)	(1, 3, 5)	(3, 5, 7)	(3, 5, 7)	(1, 3, 5)	(3, 5, 7)	(3, 5, 7) & (5, 7, 10)
Poland	(1, 3, 5) & (3, 5, 7)	(3, 5, 7)	(3, 5, 7)	(3, 5, 7)	(1, 3, 5) & (3, 5, 7)	(3, 5, 7)	(3, 5, 7) & (5, 7, 10)
Bulgaria	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(3, 5, 7) & (5, 7, 10)
Romania	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(3, 5, 7) & (5, 7, 10)
Switzerland	(7, 10, 10)	(5, 7, 10) & (7, 10, 10)	(3, 5, 7) & (5, 7, 10)	(3, 5, 7) & (5, 7, 10)	(3, 5, 7)	(3, 5, 7)	(5, 7, 10)

of Appendix B.

Step 3: Aggregate the fuzzy ratings of the fuzzy decision matrix D according to the Formula (1). The aggregated matrix of our alternatives is presented in

Table A2. Aggregated Fuzzy Decision Matrix.

Country	Cost-of-Living Index	Adjusted Quality of Life Ranking	Adjusted Adventure Ranking	Adjusted Heritage Ranking	Average Travel Cost (EUR)	Travel Time	Safety Index
Denmark	(5, 7, 10)	(7, 10, 10)	(3, 5, 7)	(3, 5, 7)	(7, 10, 10)	(3, 6, 10)	(5, 7, 10)
Austria	(3, 6, 10)	(5, 7, 10)	(5, 7, 10)	(5, 7, 10)	(1, 3, 5)	(3, 5, 7)	(5, 7, 10)
Ireland	(3, 6, 10)	(3, 5, 7)	(5, 7, 10)	(5, 7, 10)	(3, 6, 10)	(7, 10, 10)	(3, 5, 7)
France	(3, 6, 10)	(3, 5, 7)	(5, 8.5, 10)	(7, 10, 10)	(3, 5, 7)	(5, 7, 10)	(3, 5, 7)
Finland	(3, 6, 10)	(7, 10, 10)	(3, 5, 7)	(1, 3, 5)	(7, 10, 10)	(5, 7, 10)	(5, 7, 10)
The Netherlands	(3, 6, 10)	(5, 7, 10)	(5, 7, 10)	(3, 5, 7)	(5, 7, 10)	(5, 7, 10)	(5, 7, 10)
Luxembourg	(3, 6, 10)	(3, 5, 7)	(1, 3, 5)	(1, 2, 5)	(7, 10, 10)	(3, 6, 10)	(3, 6, 10)
Germany	(3, 6, 10)	(5, 7, 10)	(1, 3, 5)	(5, 7, 10)	(5, 7, 10)	(3, 5, 7)	(3, 6, 10)
United Kingdom	(3, 6, 10)	(3, 6, 10)	(1, 4, 7)	(5, 7, 10)	(1, 3, 5)	(5, 7, 10)	(3, 5, 7)
Belgium	(3, 6, 10)	(5, 7, 10)	(3, 5, 7)	(1, 4, 7)	(3, 5, 7)	(3, 6, 10)	(3, 5, 7)
Sweden	(3, 5, 7)	(7, 10, 10)	(3, 6, 10)	(3, 5, 7)	(3, 5, 7)	(7, 10, 10)	(3, 5, 7)
Italy	(3, 5, 7)	(1, 4, 7)	(7, 10, 10)	(7, 10, 10)	(1, 3, 5)	(3, 5, 7)	(3, 5, 7)
Cyprus	(3, 5, 7)	(1, 1, 3)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 4, 7)	(3, 6, 10)
Estonia	(3, 5, 7)	(1, 4, 5)	(1, 1, 3)	(1, 1, 3)	(5, 7, 10)	(7, 10, 10)	(5, 7, 10)
Slovenia	(1, 4, 7)	(1, 1, 3)	(1, 1, 3)	(1, 1, 3)	(3, 5, 7)	(7, 10, 10)	(5, 7, 10)
Latvia	(1, 4, 7)	(1, 1, 3)	(1, 1, 3)	(1, 1, 3)	(3, 6, 10)	(7, 10, 10)	(3, 6, 10)
Spain	(1, 4, 7)	(3, 5, 7)	(7, 10, 10)	(7, 10, 10)	(3, 5, 7)	(5, 7, 10)	(3, 6, 10)
Lithuania	(1, 4, 7)	(1, 1, 3)	(1, 1, 3)	(1, 1, 3)	(3, 5, 7)	(7, 10, 10)	(3, 6, 10)
Slovakia	(1, 4, 7)	(1, 2, 5)	(1, 2, 5)	(1, 2, 5)	(1, 3, 5)	(3, 5, 7)	(3, 6, 10)
Croatia	(1, 4, 7)	(1, 3, 5)	(3, 5, 7)	(3, 5, 7)	(3, 5, 7)	(7, 10, 10)	(5, 7, 10)
Portugal	(1, 4, 7)	(3, 5, 7)	(5, 8.5, 10)	(5, 8.5, 10)	(7, 10, 10)	(5, 8.5, 10)	(3, 6, 10)
Hungary	(1, 4, 7)	(1, 3, 5)	(3, 5, 7)	(3, 5, 7)	(1, 3, 5)	(3, 5, 7)	(3, 6, 10)
Poland	(1, 4, 7)	(3, 5, 7)	(3, 5, 7)	(3, 5, 7)	(1, 4, 7)	(3, 5, 7)	(3, 6, 10)
Bulgaria	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(3, 6, 10)
Romania	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(3, 6, 10)
Switzerland	(7, 10, 10)	(5, 8.5, 10)	(3, 6, 10)	(3, 6, 10)	(3, 5, 7)	(3, 5, 7)	(5, 7, 10)

in Appendix B.

$$a_{j1}=min\left\{a_{jk1}\right\},b_{j2}={\displaystyle \frac{1}{K}}\sum _{k=1}^K{b}_{jk2},c_{j3}=max\left\{c_{jk3}\right\},$$

(1)

Step 4: Determine the benefit and the cost criteria. A criterion is characterized as benefit when higher values are considered better or more desirable, and a criterion is characterized as cost when lower values are considered better or more desirable. In the presented system, the benefit criteria are adjusted quality of life ranking, adjusted adventure ranking, adjusted heritage ranking, and safety index. The cost criteria are the cost-of-living index, average travel cost, and travel time.

Step 5: Normalize the fuzzy decision matrix for alternatives Table A2 in Appendix B, according to the Formulas (2) and (3).

Table A3. Normalized Fuzzy Decision Matrix.

Country	Cost-of-Living Index	Adjusted Quality of Life Ranking	Adjusted Adventure Ranking	Adjusted Heritage Ranking	Average Travel Cost (EUR)	Travel Time	Safety Index
Denmark	(0.1, 0.14, 0.2)	(0.7, 1, 1)	(0.3, 0.5, 0.7)	(0.3, 0.5, 0.7)	(0.1, 0.1, 0.14)	(0.1, 0.17, 0.33)	(0.5, 0.7, 1)
Austria	(0.1, 0.17, 0.33)	(0.5, 0.7, 1)	(0.5, 0.7, 1)	(0.5, 0.7, 1)	(0.2, 0.33, 1)	(0.14, 0.2, 0.33)	(0.5, 0.7, 1)
Ireland	(0.1, 0.17, 0.33)	(0.3, 0.5, 0.7)	(0.5, 0.7, 1)	(0.5, 0.7, 1)	(0.1, 0.17, 0.33)	(0.1, 0.1, 0.14)	(0.3, 0.5, 0.7)
France	(0.1, 0.17, 0.33)	(3, 5, 7)	(0.5, 0.85, 1)	(0.7, 1, 1)	(0.14, 0.2, 0.33)	(0.1, 0.14, 0.2)	(0.3, 0.5, 0.7)
Finland	(0.1, 0.17, 0.33)	(0.7, 1, 1)	(0.3, 0.5, 0.7)	(0.1, 0.3, 0.5)	(0.1, 0.1, 0.14)	(0.1, 0.14, 0.2)	(0.5, 0.7, 1)
The Netherlands	(0.1, 0.17, 0.33)	(0.5, 0.7, 1)	(0.5, 0.7, 1)	(0.3, 0.5, 0.7)	(0.1, 0.14, 0.2)	(0.1, 0.14, 0.2)	(0.5, 0.7, 1)
Luxembourg	(0.1, 0.17, 0.33)	(0.3, 0.5, 0.7)	(0.1, 0.3, 0.5)	(0.1, 0.2, 0.5)	(0.1, 0.1, 0.14)	(0.1, 0.17, 0.33)	(0.3, 0.6, 1)
Germany	(0.1, 0.17, 0.33)	(0.5, 0.7, 1)	(0.1, 0.3, 0.5)	(0.5, 0.7, 1)	(0.1, 0.14, 0.2)	(0.14, 0.2, 0.33)	(0.3, 0.6, 1)
United Kingdom	(0.1, 0.17, 0.33)	(0.3, 0.6, 1)	(0.1, 0.4, 0.7)	(0.5, 0.7, 1)	(0.2, 0.33, 1)	(0.1, 0.14, 0.2)	(0.3, 0.5, 0.7)
Belgium	(0.1, 0.17, 0.33)	(0.5, 0.7, 1)	(0.3, 0.5, 0.7)	(0.1, 0.4, 0.7)	(0.14, 0.2, 0.33)	(0.1, 0.17, 0.33)	(0.3, 0.5, 0.7)
Sweden	(0.14, 0.2, 0.33)	(0.7, 1, 1)	(0.3, 0.6, 1)	(0.3, 0.5, 0.7)	(0.14, 0.2, 0.33)	(0.1, 0.1, 0.14)	(0.3, 0.5, 0.7)
Italy	(0.14, 0.2, 0.33)	(0.1, 0.4, 0.7)	(0.7, 1, 1)	(0.7, 1, 1)	(0.2, 0.33, 1)	(0.14, 0.2, 0.33)	(0.3, 0.5, 0.7)
Cyprus	(0.14, 0.2, 0.33)	(0.1, 0.1, 0.3)	(0.1, 0.3, 0.5)	(0.1, 0.3, 0.5)	(0.2, 0.33, 1)	(0.14, 0.25, 1)	(0.3, 0.6, 1)
Estonia	(0.14, 0.2, 0.33)	(0.1, 0.4, 0.5)	(0.1, 0.1, 0.3)	(0.1, 0.1, 0.3)	(0.1, 0.14, 0.2)	(0.1, 0.1, 0.14)	(0.5, 0.7, 1)
Slovenia	(0.14, 0.25, 1)	(0.1, 0.1, 0.3)	(0.1, 0.1, 0.3)	(0.1, 0.1, 0.3)	(0.14, 0.2, 0.33)	(0.1, 0.1, 0.14)	(0.5, 0.7, 1)
Latvia	(0.14, 0.25, 1)	(0.1, 0.1, 0.3)	(0.1, 0.1, 0.3)	(0.1, 0.1, 0.3)	(0.1, 0.17, 0.33)	(0.1, 0.1, 0.14)	(0.3, 0.6, 1)
Spain	(0.14, 0.25, 1)	(0.3, 0.5, 0.7)	(0.7, 1, 1)	(0.7, 1, 1)	(0.14, 0.2, 0.33)	(0.1, 0.14, 0.2)	(0.3, 0.6, 1)
Lithuania	(0.14, 0.25, 1)	(0.1, 0.1, 0.3)	(0.1, 0.1, 0.3)	(0.1, 0.1, 0.3)	(0.14, 0.2, 0.33)	(0.1, 0.1, 0.14)	(0.3, 0.6, 1)
Slovakia	(0.14, 0.25, 1)	(0.1, 0.2, 0.5)	(0.1, 0.2, 0.5)	(0.1, 0.2, 0.5)	(0.2, 0.33, 1)	(0.14, 0.2, 0.33)	(0.3, 0.6, 1)
Croatia	(0.14, 0.25, 1)	(0.1, 0.3, 0.5)	(0.3, 0.5, 0.7)	(0.3, 0.5, 0.7)	(0.14, 0.2, 0.33)	(0.1, 0.1, 0.14)	(0.5, 0.7, 1)
Portugal	(0.14, 0.25, 1)	(0.3, 0.5, 0.7)	(0.5, 0.85, 1)	(0.5, 0.85, 1)	(0.1, 0.1, 0.14)	(0.1, 0.12, 0.2)	(0.3, 0.6, 1)
Hungary	(0.14, 0.25, 1)	(0.1, 0.3, 0.5)	(0.3, 0.5, 0.7)	(0.3, 0.5, 0.7)	(0.2, 0.33, 1)	(0.14, 0.2, 0.33)	(0.3, 0.6, 1)
Poland	(0.14, 0.25, 1)	(0.3, 0.5, 0.7)	(0.3, 0.5, 0.7)	(0.3, 0.5, 0.7)	(0.14, 0.25, 1)	(0.14, 0.2, 0.33)	(0.3, 0.6, 1)
Bulgaria	(0.2, 0.33, 1)	(0.1, 0.3, 0.5)	(0.1, 0.3, 0.5)	(0.1, 0.3, 0.5)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.3, 0.6, 1)
Romania	(0.2, 0.33, 1)	(0.1, 0.3, 0.5)	(0.1, 0.3, 0.5)	(0.1, 0.3, 0.5)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.3, 0.6, 1)
Switzerland	(0.1, 0.1, 0.14)	(0.5, 0.85, 1)	(0.3, 0.6, 1)	(0.3, 0.6, 1)	(0.14, 0.2, 0.33)	(0.14, 0.2, 0.33)	(0.5, 0.7, 1)

in Appendix B is the normalized fuzzy decision matrix of our alternatives.

For benefit criteria:

$$r_{ij}=\left({\displaystyle \frac{a_{ij}}{c_j^{*}}},{\displaystyle \frac{b_{ij}}{c_j^{*}}},{\displaystyle \frac{c_{ij}}{c_j^{*}}}\right)\mathrm{and}{c}_j^{*}=\underset{i}{max}{c}_{ij}$$

(2)

For cost criteria:

$$r_{ij}=\left({\displaystyle \frac{a_j^{-}}{c_{ij}}},{\displaystyle \frac{a_j^{-}}{b_{ij}}},{\displaystyle \frac{a_j^{-}}{a_{ij}}}\right)\mathrm{and}{a}_j^{-}=\underset{i}{min}{a}_{ij}$$

(3)

Step 6: Construct a matrix A, which is a 2-dimension array with m lines and 7 columns (m are the alternative countries you want to compare, and it is lower or equal to 26 for our research). The columns correspond to the defined criteria. Each cell of A is a normalized fuzzy number (a, b, c), i.e., A is a sub-matrix of the normalized fuzzy decision matrix of Appendix B, which was constructed in step 5.

Step 7: Define for each criterion the weights of significance using one of the linguistic terms of Table 4 for each criterion. These values can be different for each individual user.

Step 8: Calculate the weighted normalized fuzzy decision matrix by multiplying each fuzzy value of matrix A with the corresponding weight (wj1, wj2, wj3) according to the following formula.

$$v_{ij}=\left(a_{ij}\ast w_{j1},b_{ij}\ast w_{j2},c_{ij}\ast w_{j3}\right)$$

(4)

Step 9: Calculate the Fuzzy Positive Ideal Solution (FPIS) and the Fuzzy Negative Ideal Solution (FNIS) according to the following formulas.

FPIS:

$$A^{+}=\left\{\langle min\left(a_{ij},b_{ij},c_{ij}\right)\mid j\in J_{-}\rangle ,\langle max\left(a_{ij},b_{ij},c_{ij}\right)\mid j\in J_{+}\rangle \right\}$$

(5)

FNIS:

$$A^{-}=\left\{\langle max\left(a_{ij},b_{ij},c_{ij}\right)\mid j\in J_{-}\rangle ,\langle min\left(a_{ij},b_{ij},c_{ij}\right)\mid j\in J_{+}\rangle \right\}$$

(6)

Step 10: Calculate the distances of each alternative from the FPIS and FNIS according to the Formula (7).

$$dd(X,Y)=\sqrt{{\displaystyle \frac{1}{3}}\left[{(a_X-a_Y)}^2+{(b_X-b_Y)}^2+{(c_X-c_Y)}^2\right]}$$

(7)

Step 11: Calculate the closeness coefficient, CCI, for each alternative to determine the ranking according to the following formula:

$$c{c}_i={\displaystyle \frac{dd(A_i,A^{-})}{dd(A_i,A^{-})+dd(A_i,A^{+})}}$$

(8)

Step 12: Rank the alternatives and select the alternative with the highest closeness coefficient.

4. Use Cases

In this section, some examples of the application of the proposed system are presented. Anna, Peter, Bill, Alice, and Kate want to travel to one of the following countries: Switzerland, Spain, Cyprus, Denmark, Germany or Sweden. Each criterion has a different importance to them. In particular, the weights of importance of each criterion for each user are presented in

Table 6. Weights of criteria for each user.

User	Cost-of-Living Index	Adjusted Quality of Life Ranking	Adjusted Adventure Ranking	Adjusted Heritage Ranking	Average Travel Cost (EUR)	Travel Time	Safety Index
Anna	VH	M	H	None	VH	None	M
Peter	L	VH	M	M	None	VL	VH
Bill	H	None	None	None	VH	VH	M
Alice	VH	H	None	H	L	None	H
Kate	L	VH	None	None	L	VH	VH
George	VH	M	VH	None	VH	None	H
Mary	M	VH	H	H	L	M	VH

. By applying the proposed fuzzy TOPSIS-based mechanism, the closeness coefficient for each alternative is calculated in

Table 7. The closeness coefficients for each alternative for use case 1.

User	Switzerland	Spain	Cyprus	Denmark	Germany	Sweden
Anna	0.28306	0.30873	0.24402	0.26545	0.24553	0.27831
Peter	0.33640	0.33335	0.23953	0.33064	0.31293	0.31289
Bill	0.20853	0.24571	0.28827	0.19456	0.20582	0.17042
Alice	0.30122	0.30182	0.23590	0.29392	0.31571	0.27209
Kate	0.31267	0.27246	0.27134	0.31837	0.29547	0.27976

. According to the closeness coefficient values, the recommended countries to travel for each user are:

• for Anna: Spain, Switzerland, Sweden, Denmark, Germany, Cyprus
• for Peter: Switzerland, Spain, Denmark, Germany, Sweden, Cyprus
• for Bill: Cyprus, Spain, Switzerland, Germany, Denmark, Sweden
• for Alice: Germany, Spain, Switzerland, Denmark, Sweden, Cyprus
• for Kate: Denmark, Switzerland, Germany, Sweden, Spain, Cyprus

The ranking of recommended destinations for each user reflects their individual criteria preferences. For example, Anna’s strong emphasis on the cost of living and adventure activities resulted in Spain ranking highest, as it aligns well with these priorities. On the other hand, Bill placed a higher importance on affordability and travel time, leading to Cyprus being his top recommendation due to its lower travel costs and convenient accessibility. Peter’s preference for safety and quality of life made Switzerland the best match for him, as it scores highly in these categories. These results demonstrate how the fuzzy TOPSIS-based mechanism dynamically balances multiple conflicting factors to generate personalized recommendations that cater to diverse traveler profiles.

Furthermore, George and Mary want to travel to a European country to ski. Their alternatives are France, Switzerland, Italy, and Bulgaria. The weights of importance of each criterion for each user are presented in Table 6. By applying the proposed fuzzy TOPSIS-based mechanism, the closeness coefficient for each alternative is calculated in

Table 8. The closeness coefficients for each alternative for Use Case 2.

User	France	Italy	Bulgaria	Switzerland
George	0.27237	0.29692	0.29988	0.28067
Mary	0.29732	0.32485	0.30253	0.26540

. According to the closeness coefficient values, the recommended countries to travel for each user are:

• for George: Italy, Switzerland, Bulgaria, France
• for Mary: Switzerland, Italy, France, Bulgaria

4.1. Results Visualization for Anna

This subsection presents a detailed example of how the proposed fuzzy TOPSIS-based system operates by focusing on Anna’s preferences. The example highlights the workflow of the system, starting from Anna’s input and preferences, followed by the computational steps, and ending with the final recommendations.

4.1.1. Anna’s Preferences and Input

Anna, as a user of the system, was tasked with selecting countries of interest and rating a set of predefined criteria that influenced her travel decisions. Anna selected the following countries as potential destinations (see Figure 1):

• Denmark
• Sweden
• Cyprus
• Spain
• Switzerland
• Germany

For each travel criterion, Anna rated the importance according to her personal preferences. The criteria and Anna’s ratings are summarized as follows (see Figure 2):

• Cost of Living: Very High (indicating that this is a critical factor for Anna)
• Quality of Life: Medium
• Adventure: High (indicating her interest in adventurous activities)
• Heritage: None (indicating no importance to cultural heritage)
• Travel Cost (EUR): Very High (indicating sensitivity to high travel costs)
• Travel Time: None (indicating no restrictions regarding travel duration)
• Safety: High (indicating her concern for safety)

4.1.2. Processing Anna’s Preferences

The fuzzy TOPSIS-based mechanism processed Anna’s preferences in conjunction with the attributes of the selected countries. Each country’s attributes (e.g., cost of living, quality of life, safety, etc.) were compared against Anna’s criteria, and the system computed a closeness coefficient for each country. This coefficient represents how closely a country aligns with Anna’s preferences.

The system then ranked the countries based on their closeness coefficients, with higher coefficients indicating a better match with Anna’s preferences.

4.1.3. Recommended Destinations for Anna

Based on the calculated closeness coefficients, the system recommended the following countries for Anna, presented in descending order of suitability (see Figure 3):

• Spain (highest alignment with Anna’s preferences)
• Switzerland
• Sweden
• Denmark
• Germany
• Cyprus

The results demonstrate the system’s ability to prioritize destinations that best meet Anna’s unique travel needs and preferences. The visual representation of Anna’s selected countries, her criteria ratings, and the ranked recommendations provide an intuitive and clear understanding of the system’s output.

4.1.4. Figures Supporting Anna’s Results

To better illustrate the process and the results, the following figures are provided:

• Figure 1: Displays the countries selected by Anna as potential travel destinations.
• Figure 2: Summarizes Anna’s criteria ratings for the travel preferences.
• Figure 3: Presents the recommended countries for Anna in ranked order.

4.2. Real-World Integration and Business Adoption

The suggested fuzzy TOPSIS-based recommendation system offers great possibilities for practical use in the travel sector. Travel platforms, online booking systems, and tourist companies could incorporate this system into their platforms to improve their personalizing ability and provide customers with more customized recommendations. Currently in use booking systems mostly suggest hotels based on user reviews, price, and location. These methods often overlook, meanwhile, subjective passenger preferences about safety, adventure, cultural heritage, or financial constraints outside of fundamental filters. Through the integration of the suggested system, these platforms could offer customized lodging recommendations based on a person’s particular travel priorities, thus improving customer satisfaction and decision-making. Similarly, travel companies and location recommendation systems might apply this method to offer customized travel plans that reflect customer preferences, financial constraints, and objectives. Exploiting our methodology would enable improved travel planning and increased customer loyalty and engagement.

Beyond accommodation and travel planning, smart city tourism applications could include this recommendation system to suggest less-crowded attractions, sustainable travel paths, or alternative off-peak travel times, assisting cities in controlling visitor flows and lower overtourism in highly sought-after locations. Using this technique, municipalities and tourism boards might highlight less well-known locations, therefore assuring more equitable distribution of the economic advantages throughout different areas.

Lastly, airlines and other transportation companies may utilize the system to maximize route and trip package recommendations depending on elements including cost-effectiveness, layovers, safety ratings, and convenience. This would enable a smoother and more user-driven travel experience in which recommendations are developed depending on actual trade-offs that guests naturally take into account.

5. Evaluation

An experiment was conducted to evaluate the proposed fuzzy logic mechanism. Specifically, a group of 27 users used the developed software, which incorporates the proposed fuzzy logic mechanism, to obtain recommendations on the best country to visit according to their preferences and needs. In addition, eight travel agency employees used the same software as a tool to generate recommendations regarding their customers’ requirements. All participants had experience with digital media and computer usage. Furthermore, before using the system, they attended a detailed demonstration. They were also provided with comprehensive instructions and user manuals. In addition, immediate support and assistance were available throughout the usage period. The characteristics of the participants are presented in

Table 9. The questionnaire for individual users.

Gender
	Male	Female
Individual users	15	12
Travel agency employees	4	4
Age
	18–25	26–35	36–45	46+
Individual users	5	10	8	4
Travel agency employees	0	3	3	2
Experience in computer usage
	None	Little	Medium	Good
Individual users	0	0	5	22
Travel agency employees	0	0	1	7
Experience in interaction with travel tools
	None	Little	Medium	Good
Individual users	0	0	9	18
Travel agency employees	0	0	1	7
Experience in interaction with ChatGPT
	None	Little	Medium	Good
Individual users	0	4	13	10
Travel agency employees	0	1	4	3

.

Then, the two groups of people complete the corresponding questionnaires that are presented in

Table 10. The questionnaire for individual users.

Questions	Content
Q1	How satisfied are you with the overall experience of using the software?
Q2	How easy was it to navigate and use the software?
Q3	Did you enjoy the overall design and appearance of the software?
Q4	How likely are you to use this software again?
Q5	Would you recommend this software to others?
Q6	Were the recommended destinations appealing and relevant?
Q7	Did the software’s recommendations align with your preferences and needs?
Q8	How would you rate the accuracy of the recommendations compared to other travel tools?
Q9	How would you rate the accuracy of the recommendations to ChatGPT’s suggestions?
Q10	Does this software, compared to other travel tools, offer a smoother and faster experience when providing recommendations?
Q11	Does this software, compared to ChatGPT, offer a smoother and faster experience when providing recommendations?

and

Table 11. The questionnaire for travel agency employees.

Questions	Content
Q1	How satisfied are you with the overall experience of using the software?
Q2	How easy was it to navigate and use the software?
Q3	Did you enjoy the overall design and appearance of the software?
Q4	Does the software save you time compared to manually researching travel recommendations?
Q5	How likely are you to incorporate this software into your daily workflow?
Q6	How likely are you to recommend this software to your clients or colleagues?
Q7	How would you rate the overall effectiveness of the travel recommendation software?
Q8	Were the recommended destinations appealing and relevant?
Q9	How relevant are the recommendations generated by the software for your clients’ preferences?
Q10	How does the design and usability of this software compare to other travel tools?
Q11	How would you rate the accuracy of the recommendations compared to other travel tools?
Q12	Does this software, compared to other travel tools, offer a smoother and faster experience when providing recommendations?

. Both questionnaires include questions that are aimed at evaluating the system’s likeability, usability, effectiveness, and accuracy of recommendations. Also, there are questions that compare the proposed system with other travel tools and ChatGPT. All questions were close-ended, with responses ranging from ‘1-Not at all’ to ‘5-Absolutely’.

The results of answering the two questionnaires are presented in Figure 4 and Figure 5. The results indicate a highly positive user experience with the software and satisfaction with the recommendations for individual users. Users are generally very pleased with their overall experience. They are satisfied with the design and appearance of the software and find it easy and straightforward to navigate. Concerning the recommendations, the users feel that they align well with their preferences and needs. The accuracy of recommendations is another strong point, with a remarkable score of 4.74 (Q8) compared to other travel tools, signaling a high level of trust in the software’s ability to offer precise travel suggestions. In addition, they believe that the proposed system provides more accurate recommendations in comparison with the AI tool ChatGPT. Furthermore, the users appreciate that the proposed software offers a smoother and faster experience than other travel tools, highlighting its efficiency and user-friendly interface. Overall, the users are very satisfied with the system’s effectiveness and usability, and they are willing to use it again and recommend it to others.

Furthermore, the experiment results reflect a highly favorable response from travel agency employees to the software. They express a strong overall positive experience and high satisfaction across multiple aspects. They find the software easy to navigate and efficient to use. Also, the system’s design and appearance are appealing to them. According to their answers, the proposed system significantly reduces the effort required compared to manual research for travel recommendations. For the above reasons, travel agency employees are willing to incorporate the proposed system into daily workflows and recommend it to colleagues or clients. Therefore, they have a strong trust and satisfaction and see the system’s value in professional settings. In addition, they are highly satisfied with the software’s ability to provide appealing and relevant destinations, and they consider that the system’s recommendations are well-aligned with clients’ preferences. When compared to other travel tools, the software excels in recommendation accuracy, simplicity, and speed.

6. Discussion

In this study, Fuzzy TOPSIS was selected as the primary decision-making approach due to its effectiveness in handling uncertainty in user preferences while providing a structured and interpretable ranking mechanism. Among various fuzzy MCDM methods, such as Fuzzy VIKOR, Fuzzy MARCOS, and Fuzzy ARAS, each has its own advantages and applications [44,45,46]. Fuzzy VIKOR is particularly useful in scenarios where compromise solutions are required, as it emphasizes minimizing the distance from an ideal solution while also considering regret-based decision-making. However, in a smart-tourism context, where the goal is to provide personalized recommendations rather than compromise solutions, Fuzzy TOPSIS offers a more straightforward ranking based on relative closeness to the ideal alternative. Similarly, Fuzzy MARCOS extends the capabilities of traditional MCDM approaches by considering both ideal and anti-ideal alternatives, which enhances decision reliability. However, it involves more computational complexity, making it less suitable for real-time personalization. Fuzzy ARAS, on the other hand, is advantageous in assessing the overall utility of alternatives but lacks the strong discriminatory power of Fuzzy TOPSIS in ranking multiple options effectively. Fuzzy TOPSIS was chosen for this study because it strikes a balance between computational efficiency, ranking precision, and robustness in handling multiple criteria with uncertainty, making it particularly well-suited for real-time and adaptive smart-tourism applications. Additionally, its ability to provide an easily interpretable ranking of alternatives based on their closeness to the ideal solution ensures a practical and user-centric recommendation system.

The findings of this study highlight the significant potential of the proposed fuzzy logic-based personalization system in enhancing the user experience within smart tourism. A key strength of the system lies in its ability to align recommendations with users’ individual preferences and needs, which is achieved through the application of fuzzy logic. By efficiently handling the inherent uncertainty and subjectivity in user data, the system enables nuanced and adaptive decision-making, ensuring precise and personalized travel suggestions that resonate with users. Tourism recommendations, which often are based on multiple criteria such as cost, travel time, activities, and quality of life, are effectively managed by Fuzzy TOPSIS, which facilitates simultaneous evaluation of these factors. Evaluation results demonstrate a highly positive response from both individual users and travel agency employees, underscoring the system’s effectiveness. Furthermore, the system’s ability to deliver a smoother and faster experience compared to other travel tools emphasizes its practical value for both personal and professional contexts. Additionally, its flexibility allows it to be applied across various tourism domains, including hotel selection, activity planning, and destination ranking, making it a scalable solution for diverse smart-tourism applications. Overall, the proposed system integrates personalization, efficiency, and usability, addressing key challenges in modern tourism and advancing the field of smart tourism.

One important consideration for real-world deployment is the system’s ability to adapt to dynamic user preferences and external factors. The fuzzy TOPSIS-based recommendation system is designed to handle real-time changes in user preferences by allowing users to update their criteria weights dynamically, ensuring that recommendations adjust accordingly. If a traveler modifies their preferences—for example, shifting priority from cost to safety—the system recalculates the rankings in real time, providing an updated list of recommended destinations.

However, at this stage, the system does not incorporate real-time external conditions such as weather, flight delays, or last-minute pricing fluctuations. While these factors significantly impact travel decisions, integrating real-time data feeds from external APIs (e.g., live weather updates or airline pricing data) would require additional development. Future iterations of the system could incorporate context-aware adaptations, ensuring that recommendations remain relevant even as external circumstances change.

While the system demonstrated significant benefits, certain limitations were identified, particularly regarding the dataset and potential biases associated with the selection of countries and subjective criteria. The present dataset is limited to particular cities and events; Athens is the main starting point, and air travel is the main means of mobility. Given that travel circumstances and choices vary greatly between areas, this choice could introduce regional and accessibility prejudices. Expanding the dataset to include more diverse destinations, multiple transport modes, and a broader range of activities will be essential for improving the generalizability and inclusivity of the system.

Additionally, the subjectivity of certain criteria must be acknowledged. The criterion “Adventure”, for example, is intrinsically interpretative since various users could relate the term to a range of activities, such as water sports, skiing, or hiking. Including a user-defined selection mechanism would help the system to increase its suggestion accuracy by letting users indicate their preferred kind of adventure experience instead of depending on a general classification.

Managing the flood of visitors to popular tourist attractions is another crucial factor, a major concern in sustainable tourism and the realm of smart tourism in general. Adding more variables, including seasonal demand swings or crowding levels, the system could propose other, less-crowded locations or off-peak travel periods, so distributing visitors in balance. This kind of approach would help a location to be long-term viable by reducing the negative effects of overtourism, improving visitor experiences, and therefore lowering pressure on local infrastructure.

Moreover, the small number of participants in the present study could limit the generalizability of the outcomes. Future studies should widen the participant pool to encompass a more varied spectrum of users from several demographic, regional, and behavioral backgrounds, strengthening the resilience and applicability of the system. This would enable a more thorough assessment of user preferences and assist in system tuning to better suit individual diversity in travel choices. This is an important avenue for future research, where a comparative evaluation with established recommendation methods will be conducted to further validate our findings and assess the system’s effectiveness against existing solutions. As the dataset scales, handling large volumes of input efficiently will become increasingly important. To address this, future research will explore optimization techniques such as parallel processing, efficient data structures, and integration with machine learning-based heuristics to enhance the scalability and computational performance of the system.

In conclusion, while the proposed fuzzy logic-based personalization system offers a promising and scalable solution for enhancing smart-tourism experiences, addressing the identified dataset limitations, subjective criteria challenges, and scalability concerns will be essential for further improving its effectiveness and ensuring broader applicability in real-world settings.

7. Conclusions

This paper presented a fuzzy logic-based personalization system designed to enhance user experience in the smart-tourism realm by generating personalized travel destination recommendations. Through the utilization of fuzzy logic in conjunction with the TOPSIS decision-making methodology, the system can successfully manage the multiple diverse, subjective, and vague data that are associated with user preferences, therefore delivering suggestions that are more accurate and relevant.

The abovementioned statement is strongly reinforced by the evaluation results, which clearly demonstrate the effectiveness of the proposed system. Both individual travelers and professionals in the tourism industry, such as travel agencies, acknowledged its value, highlighting its usability, accuracy, and the relevance of the recommendations it provides. Users found that the system significantly improved their ability to make informed travel decisions by offering travel destination suggestions tailored to their unique preferences and needs.

One of the key strengths of the proposed methodology is its ability to function both as an independent system and as a component that can be easily integrated into various tourism-related platforms. This adaptability allows it to be incorporated into online booking services, travel agency software, and personalized itinerary planners, significantly enhancing their recommendation capabilities and overall functionality. By embedding this smart-tourism system into such platforms, it can provide more accurate and user-focused suggestions, helping both casual travelers and industry professionals make better-informed decisions. Ultimately, this leads to a more efficient and enjoyable travel planning experience.

Despite its strengths, certain limitations were identified. The dataset used in the current implementation was restricted to specific cities and activities, with Athens as the primary starting point. Expanding the system to include a broader range of destinations, transport modes, and tourism categories will be crucial for its further development. Additionally, addressing tourist influx by incorporating factors such as seasonal demand and alternative destination recommendations could help manage overcrowding and improve overall travel experiences.

Future research should focus on improving the model’s adaptability, enhancing its integration with real-time data sources, and expanding the scope of evaluation to a larger and more diverse user base. By refining the system’s capabilities and addressing these challenges, the proposed fuzzy logic-based recommendation system has the potential to significantly contribute to the evolution of smart tourism, offering a robust and intelligent solution for modern travelers.