Sentiment analysis applied to tourism: exploring tourist-generated content in the case of a wellness tourism destination

ABSTRACT

Sentiment analysis of content generated by tourists through social media platforms can be used to generate knowledge about a wellness tourism destination. The present research aims to explore the content generated by tourist reviews, analyse the sentiments about a wellness tourism destination made in comments on hotels, evaluate the sentiment of tourists towards the various components of this kind of destination, and identify the motivations for tourists to look for wellness tourism destinations, such as the Algarve. The analysis of the reviews of the tourism components demonstrates tourists’ opinions about the wellness destination and also grades the infrastructure facilities and the highlighted services for the categories of this type of tourism. In attaining the objectives of this study, quantitative methods were used to review 1294 comments extracted from TripAdvisor, to which text mining algorithms were applied to assess the sentiment expressed in the comments. The results show that the three keywords most frequently used for all component categories in the Algarve wellness destinations are “great”, “love” and “visit”, and also reveal that a hotel providing spa and massage services in the tourist wellness destination mainly satisfies the desires of a tourist.

KEYWORDS:

• Tourist-generated content
• sentiment analysis
• tourism
• wellness destination
• text mining

1. Introduction

Customers of tourism products rely on social networks, and search for online reviews to obtain information on all kinds of goods, services and brands (Filieri & McLeay, 2013). A new means to quantify the in-depth quality of destinations, in terms of attractions and wellness, is through Web 2.0 applications that empower users and influence the gathering of information on travellers’ destinations (Filieri & McLeay, 2013; González-Rodríguez et al., 2016). These applications also supply credible and informative data because people share their personal travelling experience about the destination visited (Ren & Hong, 2017). Nevertheless, these are subjective opinions (positive or negative) (Vu et al., 2019).

The subjective opinion that motivates the choice of a destination can come from multiple sources: influencing agents, magazines, articles, guidebooks, word of mouth (WOM) or electronic word of mouth (e-WOM) or both (González-Rodríguez et al., 2016), and Twitter, Tumblr, and Instagram (El-Masri et al., 2017). The organic image can come through WOM provided by family and friends or individual travellers with images of past visits to the destination, e-WOM from social networks and online reviews, and information from influencing agents in published magazines, articles, guidebooks, news, films, and television promotions and documentaries by the destination management organisation (DMO) (González-Rodríguez et al., 2016; Jiang et al., 2021).

In the choice of destination, online reviews have a significant impact, but at the same time they can hinder the choice of destination because of the volume of subjective information (González-Rodríguez et al., 2016). For a tourism destination, many attractions will make the single destination attractive to tourists. The image may be the result of its natural resources, general infrastructure, tourism infrastructure, tourism leisure and recreation, culture, history and art, political and economic factors, natural environment, social environment, and atmosphere (Jiang et al., 2021).

In determining the opinion about a wellness tourism destination in a hotel review, consideration should be given to whether there is a safe and secure environment in both perception and reality, a clean and sanitary infrastructure for both locals and visitors, good quality of life for locals who benefit from tourism dollars (e.g. the creation of jobs within the industry and the creation of a market for the sale of locally produced products and services), and natural assets such as hot springs, mountains, bodies of water, forests, resources for thalassotherapy, or other natural assets, within the confines of the destination and easily accessible to visitors. Since wellness tourism and wellness travel encompass wellness for the planet, the destination must have substantive sustainability policies and practices in place, a wide range of wellness professionals and practitioners, including those offering holistic and alternative modalities, who are available and accessible a selection of hotel restaurants and independent restaurants offering a healthy cuisine prepared by chefs committed to clean eating and who work in partnership with local growers, a range of fitness-based activities and tours (e.g. yoga, hiking, cycling, fitness classes, kayaking, and stand-up paddle boarding), and a physical environment somewhat removed from the noise that has become “daily life” in the twenty-first century (Arean, 2019).

The review of the relevant literature will discuss the concepts of destination and sentiment analysis. Sentiment means the emotion and feelings (positive, neutral, or negative) of the customer. Sentiment analysis is also called opinion mining, and it studies people’s opinions about a product (Muthukrishnan et al., 2021).

This study is also important because it will be a guide both to public policy managers on making and executing policy to further improve the sustainability of destinations and improve the patronage of tourism places, and also to the private sector, who are usually in direct contact with tourists who visit wellness destinations.

2. Literature review

2.1. Tourism destinations

A destination is a place where someone is going or where something is being sent or taken, or somewhere worth making a special journey to. The word is used with certain other words to make phrases, including “attractive destination” and “favourite destination”. When the physical place and its supporting amenities, facilities, activities, interests, and attractions are combined, the resulting geographical space is called a tourist destination (Klein-Hewett, 2021). For Komilova et al. (2021), a tourism destination is a geographical region with a particular attractiveness for tourists that meets the following basic requirements:

1. Availability of the necessary service complex for the reception of tourists;

2. Availability of attractive objects and places of interest to tourists; and

3. Availability of information systems.

Żemła (2016) explained the concept of a tourism destination using five approaches: the spatial approach, the economic approach, the managerial approach, the systems approach, and the network approach.

In the spatial approach, a tourism destination is considered to be a geographical area or region in which the visitor enjoys various types of travel experiences through physical entities and intangible socio-cultural entities which motivate tourists to visit and encourage tourism activities (Żemła, 2016).

The economic approach considered by Żemła (2016) is divided into the demand side and the supply side approach. The demand side approach defines a tourism destination as a choice made by a tourist that ensures the satisfaction of their special holiday needs in the areas of accommodation, catering and entertainment, cultural background, purpose of visit, educational level, past experience and travel itinerary, although Krakover and Corsale (2021) identified time and money budgets as a constraint on the repetition of a trip. The supply-side approach defines a tourism destination as an area, country, region or human settlement with a significant offer of attractions and tourism infrastructure for visitor experiences, containing a concentration of small and medium-sized companies that offer tourism products or services.

The managerial approach to the concept of destination, according to Żemła (2016), considers a destination as a product of an area with a different combination of tourism products and services that provide a unique experience. For a company, these may be seen as strategic business units from the management point of view that are providing products and offers for defined target groups and guest segments.

According to Żemła (2016), the system approach definition considers a destination as a place in which actors cooperate in order to supply an integrated tourist product. The actors are from the private and the public sector.

For the network approach, Żemła (2016) stated that a tourism destination is a complex system, an area bound by no administrative limitations, in which tourism aspects are interrelated and integrated.

After a review of the various conceptualisations of destinations, Jovicic (2019) focused on three concepts of a destination: the classical – traditional view of a destination, the systemic approach to a destination, and the concept of a smart tourism destination.

Using the classical – traditional view of a destination, Jovicic (2019) defined a tourist destination as a destination area in which tourism occurs as an industry – an area with different natural and/or man-made features that attracts non-local visitors (or tourists) for a variety of activities, meeting criteria such as having tourist attractions and accommodation and good transportation to, from, and within the destination. Jovicic did not emphasise co-operation within the destination or the role of tourists as actors in the destination. His view sees tourists only as consumers of a destination’s supply of services, ignoring the fact that change in demand leads to changes in the structure of the destination. The systemic approach to destinations, according to Jovicic (2019), considers a tourism destination in the context of other systems that interact with tourism. It treats a destination as an open and flexible system that is characterised by a high degree of interaction between its constituent elements, such as firms providing tourist services, residents in the destination area, local authorities and tourists, and also as a network of connected organisations and stakeholders whose productivity is essential for the functioning of the destination system.

For Jovicic (2019), a Smart Tourism Destination is a knowledge-based destination where information and knowledge related to tourism are instantly exchanged using information and communication technologies (ICTs). In this environment, the DMOs by actively listening to tourists’ opinions, suggestions, and feedback; can better meet their needs and transform tourists into co-creators of the destination’s offerings. As ICTs continue to advance, increased communication and collaboration between tourists and destination stakeholders are expected, leading to high-quality tourist experiences and successful business outcomes.

2.2. Wellness tourism

2.2.1. The components that characterise a wellness tourism destination

Kathleen Lesage (Wellness Tourism Association, 2022) defined wellness tourism as tourism that offers personal growth, outdoor adventures, nutritious cuisine, spa treatments, bodywork and opportunities in preventative care, all while on vacation. Wellness has also been defined as the sum of all the relationships resulting from a journey by a tourist whose primary motive is to maintain or promote their health and well-being and who stays at least one night at a facility that is specifically designed to enable and enhance physical, psychological, spiritual and/or social well-being (Stará & Peterson, 2017; Valentine, 2016). Thal and Hudson’s (2019) study emphasised that, in enhancing psychological, social well-being, the quality and standard of the service-scape and the relationship between staff and guests are most important for creating a psychologically beneficial environment, so wellness tourism is related to travelling for maintaining well-being and improving lifestyle (Han et al., 2020). The term “wellness” also refers to “a special state of health, incorporating the harmony of body, mind, and spirit” and one’s lifestyle (Han et al., 2020, p. 425).

Also, Pan et al. (2019) defined wellness as the state of optimal physical, mental, and social well-being but also, with the new trend in the market due to the boom in wellness tourism, with many holistic and individual perspectives. Wellness tourism also covers some specific activities, which include personal service, health promotion treatments, the environment, a healthy diet, relaxation, social activities, the experience of unique tourism resources, and mental learning (Pan et al., 2019). The holistic perspective of Dini and Pencarelli (2021) provides a more detailed list of the wellness activities that make up each component of wellness tourism: hot springs, spas, care of body and mind, medical tourism, the natural environment, spirituality, culture, enogastronomy, sports, and events.

Wellness tourism relates to the adoption of a healthy way of life. Because of the effects of the economy and globalisation, this implies an expanding tourism niche encompassing individual or group travel to specialised resorts and destinations for the purpose of maintaining physical and mental health. Kazakov and Oyner (2021) have ten categories of wellness tourism, and words associated with each category, which are set out in Table 1.

Table 1. Wellness tourism dimensions/categories and words associated.

Categories	Words associated	Sources
Hot springs	Mineral and thermal waters; hot springs establishments/hotel; thermal springs/centre/ complex/; health preservation bath	Dini and Pencarelli (2021); Rodrigues et al. (2020); Brandão et al. (2021).
Spas	Hotels-spa; spas; spa tourism; spas destination; luxury spa	Lemi (2021); Brandão et al. (2021); Dini and Pencarelli (2021).
Care of body and mind	Sauna and beauty service; fitness; cosmetics; massages; rhythmic movement therapy (RMT); gyms	Dini and Pencarelli (2021); Plzáková and Stupková (2019).
Medical tourism	Medical tourism; healthy; tourists with cancer and other chronic diseases; aged nursing; babymoon destination; healthcare and wellness tourism	Plzáková and Stupková (2019); Brandão et al. (2021); Dini and Pencarelli (2021).
Natural environment	Nature; environment; rural; therapeutic landscapes; social value; coastal areas; lakes; adventure tourism/ ecotourism; arboretum	Dini and Pencarelli (2021).
Spirituality	Spiritual; spiritual revolution; yoga; new age	Dini and Pencarelli (2021); Zhang et al. (2021); Han et al. (2020); Pan et al. (2019); Plzáková and Stupková (2019); Stará and Peterson (2017); Valentine (2016).
Culture	China wellness culture; longevity in Chinese culture; local culture; local way of life.	Dini and Pencarelli (2021).
Enogastronomy	Food quality and/or food service fresh and local; nutritional foods; healthy food; indigenous food; dietary	Madeira et al. (2019); Dini and Pencarelli (2021).
Sports	Sports tourism; sports and recovery; horse riding; golf	Dini and Pencarelli (2021).
Events	Festival and events; savonlinna opera festival	Dini and Pencarelli (2021).

Source: Adapted from Dini and Pencarelli (2021, pp. 339–340).

For those making travel decisions and those tracking tourist behaviour, user-generated content (UGC), including narrative text and photographs, is a powerful platform allowing potential tourists to collect travel information and, also, an essential channel for destination image perceptions (Huang et al., 2022).

2.2.2. Sentiment analysis in user-generated content (UGC)

Kirilenko et al. (2017, pp. 1013–1014) define sentiment analysis or opinion mining as “the computational study of people’s opinions, appraisals, attitudes, and emotions toward entities, individuals, issues, events, topics and their attributes’. Sentiment analysis identifies what people like and dislike and helps build recommendation systems and targeted marketing campaigns (Muthukrishnan et al., 2021). There are four types of sentiment analysis (Muthukrishnan et al., 2021): fine-grained sentiment analysis, emotion detection sentiment analysis, aspect-based sentiment analysis and intent sentiment analysis.

The sentiment classification of the identified aspects, as presented in Figure 1, has to do with the classification of the multiple aspects of opinions gathered from online reviews, Twitter, Facebook and other blogs that provide reviews of wellness destinations.

Figure 1. Process of opinion mining in multiple aspects.

Source: Muthukrishnan et al. (2021, p. 5185).

Description of the opinion mining process, from the collection of reviews, which convey the opinion of tourists, then the features are extracted to analyse the extracted phrases and obtain the result.

By understanding the mindset of a customer about its products or services and the intention of a customer to contact the company about those products or services, a company can guess whether or not a customer intends to use a product. This means that the intentions of that exact customer can be traced, forming a pattern which is used for marketing and advertising (Muthukrishnan et al., 2021), as presented in Figure 2.

Figure 2. Semantic pattern model. Source: Rodrigues et al. (2020, p. 652).

Description of the semantic pattern model, from word analysis to obtaining associated themes.

To identify the key drivers of wellness tourism satisfaction and dissatisfaction, Rodrigues et al. (2020) used software called Leximancer featuring a “Sentiment Lens”. Sentiment Lens only applies sentiment terms that identify words that reveal favourable sentiment terms and unfavourable (or negative) sentiment terms. Also, sentiment analysis can provide further insights by organising tourists’ descriptions in terms of negative or positive appraisals as relevant and as used consistently within all the reviews (Rodrigues et al., 2020).

Sentiment analysis, therefore, seeks to determine the attitude of speakers or writers towards some topic, or the overall contextual polarity of documents (Rodrigues et al., 2020).

2.2.3. Present study and conceptual research model

The present study aims to explore the tourist UGC as related to Table 1 in the case of a wellness tourism destination and to use sentiment analysis to bring together the various views on wellness tourism.

The sentiment analysis of the reviewed text has a single mode because it involves text extracted from only a single online review source (TripAdvisor). Machine learning is used for the analysis, as illustrated in Figure 3.

Figure 3. Sentiment analysis process on product reviews.

Source: Adapted from Medhat et al. (2014, p. 1094).

Description of the sentiment analysis process on product reviews, which ends with obtaining the polarity associated with the sentiment.

The main aim of exploring tourist-generated content is to analyse the feeling about wellness tourism destinations in the reviews of hotels, to evaluate the sentiment of tourists towards the components that characterise a wellness tourism destination, and to identify the motives that lead tourists to look for wellness tourism destinations in the Algarve. A total of 1294 comments were extracted from TripAdvisor to achieve these aims. To measure the sentiment expressed in the comments, text mining algorithms were applied.

3. Research methodology

There are five main steps in performing sentiment analysis (El-Masri et al., 2017): data collection, pre-processing, feature extraction, application of the sentiment analysis algorithm, and evaluation of the results.

The first objective of the research is to analyse the sentiments about wellness tourism destinations in TripAdvisor reviews, and total of 1294 reviews from destinations in the Algarve were manually extracted, as presented in Table 2. The entities were chosen because of their review ratings, services, serene environment and additional tourist services provided by complementary tourist organisations.

Table 2. Algarve destination TripAdvisor.

ID	Name	City	Country	Category	Star	No. of Reviews
1	Caldas de Monchique	Faro	Portugal	Hot springs	4.0	156
2	Zensation – Body & Soul	Albufeira	Portugal	Care of body and mind	4.5	142
3	Mercado de Loulé	Loule	Portugal	Enogastronomy	4.5	160
4	Moinhos Velhos Juice Detox & Yoga Retreat	Lagos	Portugal	Spirituality	4.5	50
5	Old Town Faro	Faro	Portugal	Culture	4.0	160
6	Pinetrees Horse Riding	Almancil	Portugal	Sport	5.0	156
7	Ria Formosa Natural Park	Abufeira	Portugal	Natural environment	4.5	160
8	Swan Day Spa	Alvor	Portugal	Spas	5.0	160
9	Zoomarine Algarve	Albufeira	Portugal	Event	4.5	150
Total						1294

Source: Author’s elaboration.

Since the study context was applied to the Algarve, where a given typology only presents one entity, only one entity was considered for each of the Wellness tourism dimensions shown in Table 1 (Mountassir et al., 2012). This is because when applying sentiment analysis, it is crucial that the data consider a balanced sample. For instance, as shown in Table 2, no representative entity with TripAdvisor ratings was found, hence medical tourism was not examined.

The pre-processing in the text mining involved: transformation, tokenisation, normalisation and filtration. The words, especially words in a phrase or a sentence, were transformed into lowercase. Tokenisation of the full text was carried out by breaking up the stream of words, phrases, and symbols and removing the punctuation (El-Masri et al., 2017; Vishwanathan, 2010). Stop words, numbers and regular expressions were filtered, and the text was normalised by removing prefixes and suffixes in word sentences.

The pre-processing of data from the corpus led to the extraction (see Table 3) of keywords. It also identified keywords for each of the nine categories in the Algarve destinations. Since there were no online reviews for the medical tourism category, it was therefore not considered in the analysis.

Table 3. Popular hotels and its category visited by tourists.

Entity	Caldas de Monchique	Swan Day Spa	Zensation – Body & Soul	Ria Formosa Natural Park	Moinhos Velhos Juice Detox & Yoga Retreat	Old Town Faro	Mercado de Loulé	Pinetrees Horse Riding	Zoomarine Algarve
Category	Hot springs	Spas	Care of body and mind	Natural environment	Spirituality	Culture	Enogastronomy	Sport	Event
Thermal	11
Health	30
Spa		248
Luxury		4
Massage			199
Sauna			18
Healthy
Medical
Nature				55
Landscape				9
Spiritual					9
Yoga					35
Culture						8
Wellness						2
food							127
Nutrition							9
Sport								6
Golf								5
Festival									4
Event									9
Count	41	252	217	64	44	10	136	11	13

Source: Author’s elaboration.

The data which represents the identifier of each component category, is designated by ID, Hot springs, Spas, Care of body and mind, Medical tourism, Natural environment, Spirituality, Culture, Enogastronomy, Sports, and Events, although no reviews were generated for Medical tourism.

The Vader algorithm (Hutto & Gilbert, 2014) was taken into consideration as a technique for sentiment analysis. The results of this algorithm range from −1 (most negative sentiment) to +1 (most positive sentiment), i.e. it presents a positive score, negative score, neutral score, and compound (combined score) for each comment. To meet the objectives of the current study, which are to determine the factors that positively motivate destination choice and, conversely, the negative emotions that may motivate non-choice, only the positive, negative, and combined scores will be considered. As a result, the neutral score was not taken into account during the analysis that was conducted.

4. Findings and discussion

4.1. Description of data and sample characteristics

The empirical analysis in this research employs reviews on TripAdvisor of Algarve destinations. The destination hotels were categorised according to the ten categories and the words associated with each category, as presented in Table 1, and a total of 1294 reviews were manually extracted from TripAdvisor. Only nine of the categories were used as there were no reviews related to medical tourism within the period of the reviews collected by the researcher, which is shown in Table 2.

The objectives of the research are as follows:

1. To analyse the sentiments about wellness tourism destinations found in the reviews of the entities.

2. To evaluate the opinions of the tourists based on their destination reviews; this is subdivided into specific objectives:

   • Identify the components that characterise a wellness tourism destination.

   • Collect the comments posted on the social network to form the sample.

   • Apply the different pre-processing (cleaning) techniques to the database.

   • Apply text mining algorithms to assess the sentiment expressed in the comments.

   • Considering the sentiment analysis results, identify the motivations that lead tourists to choose wellness tourism destinations in the Algarve.

The feelings about the wellness tourism destinations in the reviews were analysed using the information in Figures 4 and 5 and Tables 2 and 4, which present the sentiments from the full text. Also, the opinions based on the destination reviews can be evaluated by looking at Figures 4–13, and also considering the content of Table 3.

Figure 4. Sentiment by author country.

Source: Author’s elaboration.

Presents the average review rating compared to the average normalised sentiment (on a scale of 1–5) for each country of review authors.

Figure 5. Sentiment by Entity.

Source: Author’s elaboration.

It presents the average review rating compared to the average normalised sentiment (on a scale of 1–5) by entity that represents each type of health and well-being tourism considered in the study.

Figure 6. A word cloud of all comments.

Source: Author’s elaboration.

It presents the word cloud of all comments where the words “love”, “great”, “visit”, “ride,” and “place” stand out.

Figure 7. Destination Sentiments versus year with winters effect.

Source: Author’s elaboration.

It presents the average review rating compared to the average normalised sentiment (on a scale of 1–5) over the months of the year.

Figure 8. Sentiment with Covid effects.

Source: Author’s elaboration.

It presents the average review rating compared to the average normalised sentiment (on a scale of 1–5) over the year, highlighting the effect of the economic crisis in 2010 and COVID in 2021.

Figure 9. Author country versus sentiment.

Source: Author’s elaboration.

Presents the sentiment for each country of review authors considered in the study.

Figure 10. Number of the reviews by entities.

Source: Author’s elaboration.

Presents the number of reviews per entity from 2009 to 2022.

Figure 11. Average of sentiment and average of review ratings by wellness category.

Source: Author’s elaboration.

Compare the average of sentiment and average of review ratings by wellness category.

Figure 12. Average of sentiment and average of review ratings by words of wellness category.

Source: Author’s elaboration.

Compare the average of sentiment and the average of review ratings by the words associated with the wellness category.

Figure 13. Average of sentiment and average review ratings by country of the author’s review considered by the wellness category.

Source: Author’s elaboration.

Figure 4 shows the relationship between the average review rating and the average sentiment, normalised by the author’s country, associated with all the comments collected from TripAdvisor. To carry out the normalisation, the formula presented by Chen et al. (2020, p. 9) was considered, and presented in equation (1): (1) $x_i^{{}^{\mathrm{\prime }}}={\displaystyle \frac{x_i-x_{\min }}{x_{\max }-x_{\min }}}$(1) Figure 5 shows the relationship between the average review rating and the average sentiment, normalised by the entities, associated with all the comments collected from TripAdvisor and Figure 6 presents the words that appear most often in the reviews.

In general, and taking into consideration the word cloud presented in Figure 6, the words that are most commonly used to represent the Algarve as a wellness destination are “love”, “great”, “visit”, “ride” and “place”.

4.2. Sentiment score by wellness tourism category

Taking into consideration the categories presented in Table 2, and the data collected, it is possible to create Table 4, where the sentiment is presented for each wellness tourism category, and Table 5, the wellness tourism category and the keywords for the destination.

Table 4. Sentiment comments and keyword analysis about the wellness tourism destination.

ID	Average ReviewRating	Average of sentiment normalised (1–5)	Sentiment Average (−1 to +1)	Keywords of positive comments	Positive words most used in negative sentiments	Negative words most used in negative sentiments
1	3.94	4.29	0.67	Nice Place spa	Spa Place Nice	Place Spa water
2	4.97	4.94	0.93	Massage Spa Relax	Massage Spa Relax	Massage Spa Relax
3	4.73	4.66	0.82	Fish Massage Spa	Fish Massage Spa	Massage Back time
4	4.35	4.25	0.67	Park Bird Walk	Park Bird Walk	See Walk Visit
5	4.66	4.76	0.89	Experience Time Great	Experience Great Place	Staff Time
6	4.28	4.28	0.67	Old Town Faro	Old Town Faro	Town Old Faro
7	4.77	4.83	0.90	Ride Horse Pinetree	Ride Horse Pinetree	Day Ride Park
8	3.79	4.03	0.59	Day Ride Park	Day Ride Great	Ride People Park
9	4.3	4.51	0.75	Market Fish Visit	Market Fish Visit	Market Visit Local
Total	4.40	4..48	0.76	Visit love great	Love Great Visit	Time Place Visit

Source: Author’s elaboration.

Table 5. Wellness tourism category and keywords about wellness tourism destination.

Category	Features	No of reviews	Minimum	maximum	Average	Standard Deviation
Hot springs	Thermal health	41	0.3	5	4.5	1.1
Spas	spa luxury	251	0.6	5	4.6	0.8
Care of body and mind	massage sauna	210	0.2	5	4.7	0.6
Natural environment	nature landscape	63	2.7	5	4.7	0.4
Spirituality	spiritual yoga	36	0.1	5	4.7	1.1
Culture	culture welness	10	0.3	5	4.1	1.7
Enogastronomy	food nutrition	135	0.1	5	4.3	1.1
Sport	sport golf	11	4.2	5	4.7	0.3
Event	festival event	13	0.4	5	3.2	1.9

Source: Author’s elaboration.

Table 4 illustrates the statistics for the keywords related to positive comments, the positive words most used in negative sentiments, and the negative words most used in negative sentiments for the wellness destinations.

For ID1, most tourists describe the destination as a nice place, and they say that the spa facilities are also nice, but are nothing special. Some of the few negative sentiments are about the issue of the spring not flowing.

For ID2, the keywords are massage, spa and good for relaxing.

For ID3, the positive comment on the destination is about the unique fish massage spa, and the negative comment is about the timing of the pedicure and massage.

The impression of the tourists for ID4 is that the entity are just like a garden, with the sound of birds, and are also a nice place for relaxation, but that there is not much to see or visit at the destination. The tourist’s experience was great, but there are negative comments on the attitude of the staff at ID5 and also about creating time for another visit.

For ID6, positive keywords about the old town of Faro are frequently used. The tourist in ID7 enjoyed horse riding among Pine trees.

In ID8, the entity provided a satisfying day’s ride, but there were too many people every day to ride in the park. For ID = 9, the negative comments are that the market was local, but the reviews say that it is a market for different types of fresh fish, and is worth visiting.

Table 5, showing average ratings of 4.6 and 4.7 and standard deviations of 0.8 and 0.6, respectively, for Spas and Care of body and mind, shows that a lot of sentiment was attached to spas, luxury, massages and saunas. This clearly expresses the acceptability and popularity among tourists of wellness destinations with Spas and Care for body and mind. This aspect is analysed by Suban (2024), who states that managers of tourist destinations should be aware of how essential satisfaction is in motivating visitors’ behavioural intentions, which highlights the importance of spa tourism from a wellness perspective.

Also, Natural environment and Sport, with an average of 4.7 and standard deviations of 0.4 and 0.3, show that wellness destinations with sport (golf), nature and landscape are acceptable to tourists.

The authors’ sentiments from the Full text are presented in Figures 4 and 5, with a total of 1294 comments, as shown in the review of the comments in Table 4. The positive comments make up 24% of the total, while 3% of the sentiment comments are negative. The destinations ID1, ID2, ID3, ID6, ID7 and ID8 have more “excellent” reviews, but this might be because they had good service previously, as the change in service experience due to Covid influences the sentiment review rating vis-à-vis the average positive and negative sentiments (Figures 5 and 8).

Comparing the ReviewRating, which is the assessment attributed by the tourist, with positive and negative sentiments, Figure 5 illustrates that the ReviewRatings are greater than the positive and negative sentiments. Also, comparing the normalised sentiment with the positive and negative sentiment, Figure 10 illustrates the hotels’ yearly average reviews, with Zoomarine Algarve having the highest reviews between 2019 and 2022. The reviews were at their highest during the winter months of November, December, January, February, and March (Figure 7).

Table 3 illustrates the opinions of the tourists on each of the hotels. There are 41 words in the Hot springs category, 252 in the Spas category, 217 in the Care of body and mind category, 64 in the Natural environment category, 44 in the Spirituality category, 10 in the Culture category, 136 in the Enogastronomy category, 11 in the Sport category, and 13 in the Event category.

Table 4 identifies the term frequency – inverse document frequency (TF-IDF) of the motivating words used by the reviewers of the nine entities in the Algarve. In all the reviews of the nine entities, the reviewers most frequently used the words great, love and visit. These high TF-IDF words are motivating words which explicitly express positive sentiments about the Algarve destination.

When considering the keywords in the positive comments of the reviewers (Table 4), one should bear in mind (Rodrigues et al., 2020) that writers and speakers on some topics and attitudes are determined by sentiment. Also, according to Żemła’s (2016) and Jovicic’s (2019) definitions of a tourist destination, the Algarve destination has natural and man-made features which attract non-local visitors, as illustrated in Table 3 The destination has helped visitors to maintain their well-being and improve their lifestyle (Han et al., 2020) and, in support of the view of Thal and Hudson (2019), the reviews show that the relationship between staff and guests has created a psychologically beneficial environment and also optimal physical, mental, and social well-being (Pan et al., 2019). The keywords in the table for positive comments include “nice”, “place”, “great”, “experience”, and so on, which simply means that the Algarve wellness destinations are worth visiting again. Some demotivating factors were discovered when manually extracting the reviews; these include hidden charges, time spent in accessing the destination area or the tourism facilities due to high volume of patronage at a particular time, the attitudes of staff at the destination, selective service delivery, the absence of standard rules, and a lack of correlation between the service advertised and the service delivered. According to Krakover and Corsale (2021), these problems affect tourists’ time and money budgets and also discourage them from visiting the wellness destination again.

As shown in Figure 7, the average for the sentiments is higher in the summer months, when compared with the average rating.

Figure 8 shows the average for the sentiment and ratings over the years, and it can be seen that during the pandemic period the average sentiments and ratings dropped, because of the feeling of insecurity experienced by tourists. From the summer of 2021, both values rose again, but they remained lower than the pre-pandemic period values. In Figure 8, one can still analyse the emergence of the economic crisis of 2008, whose effect was more pronounced in 2010 (FFMS, 2024), mainly affecting European countries. In 2009, world GDP fell by 0.6%, with the first global recession occurring after World War II, which resulted in discontent and dissatisfaction among the population (Supiot, 2010), reflected in their sentiments.

In Figure 9, it can be seen that tourists from Ireland give higher ratings and sentiments while, on the other hand, those from France give the lowest values, in terms of ratings, followed by those from the Netherlands. However, for Belgian tourists the average sentiment is high and there is a marked distance from the ratings.

Figure 10 presents the number of reviews for each entity. It can be seen that this number has grown, but suffered a slight decrease in the pandemic period. In other words, more and more tourists are sharing their experience on social networks.

Taking into account the wellness destinations, Table 6 presents the average of the sentiment and rating associated with each category defined in Table 2 for the Algarve destinations. This is complemented by the graph in Figure 11, which highlights that the “Care of body and mind” and “Natural environment” categories were ranked with the highest sentiments, followed closely by the “Spa”, “Spirituality” and “Sports” categories. The “Events” category has the lowest rating.

Table 6. Average of sentiment and average of review ratings by wellness category.

Category	Average of sentiment	Average of ReviewRating
Hot Springs	4,6	4,1
Spas	4,7	4,6
Care of Body and mind	4,8	4,9
Natural Environment	4,8	4,6
Spirituality	4,7	4,6
Culture	4,2	4,3
Enogastronomy	4,4	4,2
Sport	4,7	4,5
Event	3,2	3,2

Source: Author’s elaboration.

Regarding the average ratings, the category “Care of body and mind” has the highest average, and “Events” the lowest. For Events, the evaluation for the sentiment and the rating level is the same but, on the contrary, “Hot springs” has a large difference between the sentiment and the rating, which may mean that it helps the tourists’ well-being.

In terms of the keywords considered for each category of wellness destination, Table 7 presents the average of the sentiment and the review ratings given by the tourists. This is complemented by the graph in Figure 12, which highlights the fact that the experiences associated with the luxurious environment of Spas, as in the well-being provided by Culture and the nutritional part of the Oenogastronomy category have the highest value in terms of sentiment and review rating, with the exception of Nutritional experience which has a low review rating value when compared to sentiment. This insight is in line with the work of Xia et al. (2024), which states that tourists who value the experience and quality of the spa most are more satisfied with their tourism experience, and the work of Roy (2023) which concludes “health and wellbeing” have become one of the most prominent factors of emotional aspects leading to a satisfying tourism experience.

Table 7. Average of sentiment and average of review ratings by words considered in the wellness category.

	Word	Average of sentiment	Average of ReviewRating
Hot Springs	Thermal	4,5	3,7
	Health	4,6	4,3
Spas	Spa	4,7	4,6
	Luxury	5,0	5,0
Care of Body and mind	Massage	4,8	4,9
	Sauna	4,9	4,8
Natural Environment	Nature	4,8	4,6
	Landscape	4,7	4,9
Spirituality	Yoga	4,7	4,6
	Spiritual	4,4	4,4
Culture	Culture	4,0	4,1
	Welness	5,0	5,0
Enogastronomy	Food	4,4	4,2
	Nutrition	5,0	4,4
Sport	Sport	4,7	4,2
	Golf	4,8	4,8
Event	Festival	4,5	4,5
	Event	2,7	2,7

Source: Author’s elaboration.

In the analysis of Table 7 and Figure 12, it should be highlighted that the experience associated with Events arouses negative sentiment, so this should be a category that Algarve destination managers pay attention to in the future.

In terms of countries of origin of tourists and the analysis of sentiment, the Portuguese themselves stand out as negatively evaluating the Algarve in terms of Events, followed by Spain and the United Kingdom. The category of Spirituality obtained the maximum score in the feelings expressed by tourists, as did the Hot springs category in almost all the countries evaluated, except the United Kingdom, as presented in Table 8 and Figure 13.

Table 8. Average of sentiment by country of the author’s review considered by the wellness category.

	Belgium	Canada	France	Germany	Ireland	Netherlands	Other	Portugal	Spain	UK	Unknown	USA
Hot Springs	5,0				5,0	5,0	5,0	5,0		4,3	5,0	5,0
Spas	4,0	4,0	4,0	4,2	5,0	4,0	4,9	4,6	5,0	4,6	4,9	4,5
Care of Body and mind	5,0	5,0	5,0	4,5	5,0	5,0	4,3	4,9	5,0	4,8	4,9	4,5
Natural Environment	5,0	5,0	4,0	4,0	5,0	4,3	4,7	5,0		4,8	5,0	5,0
Spirituality	5,0						5,0	5,0		4,5	5,0	5,0
Culture	5,0							4,0		3,3	5,0
Enogastronomy		5,0	5,0	4,5	4,0	4,5	3,7	4,7	5,0	4,4	4,5	4,3
Sports					4,0	5,0		5,0		4,8
Event	5,0						5,0	1,0	3,0	3,0	2,7	5,0

Source: Author’s elaboration.

5. Conclusions

5.1. Practical implications for the management of the tourism sector

Sentiment analysis applied to tourism allows us to explore the opinion of tourists from the content they generate. In essence, it helps to stimulate opinion on destinations with wellness entities, and reveals how satisfied individuals generally are with tourism wellness destinations based on keywords that illustrate their opinions. It also widens the knowledge of the DMO about the sustainability of the destination infrastructure arising from tourists’ views and opinions on the destination. However, sentiment analysis of autonomous generated content still has some implications for the management of the tourism sector.

The generated content of a review may depend on the psychological timeframe of the individual reviewer, which is subject to change as a result of implicit factors. These factors may be the general mood of the reviewer when making a judgment about the services received, meaning that their feedback becomes ambiguous and does not help in the growth of the sector.

The Full text, the ReviewRating and the sentiment analysis results are sometimes not congruent. For example, for the Full text “Bring your hiking shoes if visiting this area. Plan to stay for a few days at least, this was our biggest regret that we were not staying longer”, the ReviewRating is 5, the sentiment normalisation is 1 and the sentiment opinion is negative.

As another example, the Full text was: “We had a very enjoyable holiday in a bubble resort in the Algarve. But that is the best way to experience the area. Venture outside your manicured paradise and you’ll find a fairly characterless, soulless place, typified by the “historic” old town in Faro. This was really only a collection of cafes and restaurants in a block or two of cobbled streets where cars freely drove. It was an endurance test for our family and toddler. But it was also a reminder that this is holiday resort land and not full of beautiful, picturesque towns. If you’re thinking about visiting, don’t”. Here the ReviewRating is 1, the sentiment normalisation is 5 and the sentiment opinion is positive.

Although the majority of the Full text was congruent with the ReviewRating, sentiment normalisation and sentiment opinion, the outliers can still have effects for the management of the tourist sector.

5.2. Contributions to the scientific community

The contributions of this research study are in the areas of theory, methodology and practice. The theoretical study examined tourist destinations as an entity and the characteristics of wellness tourism destinations. The components of wellness destinations (Dini & Pencarelli, 2021) were theoretically examined, building on other studies. Unlike earlier studies that give a descriptive analysis deduced from other literature, this study applied the scientific method through data mining to analyse the sentiment of tourists toward each of the components in the wellness destinations. The sentiment analysis used a corpus graph to extract the most common keywords (visit, great, and love) for all the component categories.

The study using the sentiment analysis explored the wellness categories in the destinations most visited by tourists, showing the economic viability of those categories in the management of the tourist sector as another contributor to gross domestic product. It also showed the weakness of other categories, and these will need attention from the management of the tourist sector.

5.3. Limitations and future research

Social networks and online reviews are sources of information for tourism and all kinds of goods and services (Filieri & McLeay, 2013). They are also sources of credible and informative data because the personal experiences of travellers are shared for public consumption (Ren & Hong, 2017).

In the analysis of the results using text mining algorithms to assess the sentiment expressed in the comments, the three keywords most frequently used in all categories of the wellness components in relation to Algarve wellness destinations are identified in Table 4. The opinions on the popular hotels and the wellness categories visited by tourists in the Algarve wellness destinations are illustrated in Table 3, and all the graphs illustrate the relationship between ReviewRating and the normalised sentiment.

The study was limited to the Algarve region in Portugal, which includes Faro, Albufeira, Loule, Monchique, and Olhão. The supposed components are ten, namely: Hot springs, Spas, Care of body and mind, Medical tourism, Natural environment, Spirituality, Culture, Enogastronomy, Sports, and Events. When the reviews were manually extracted from TripAdvisor on the Algarve wellness destinations, there was no review on Medical tourism. However, it is suggested that, for further study, the process of opinion mining be carried out in multiple locations, such as reviews, Twitter, Facebook (Muthukrishnan et al., 2021) and Google review, so that there will be wider sources of information, especially as regards Medical tourism.

We also want to analyse the Algarve region as a “smart tourism destination” and its potential for brand development and creation as a “smart wellbeing tourist destination” in addition to this research path. It will be supplemented by a second study that examines the adoption of particular biometric data collection technologies and their impact on improving the overall quality of the travel experience in terms of emotions, physical well-being, and mental health.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This paper is financed by National Funds provided by FCT- Foundation for Science and Technology through project UIDB/04020/2020 and with DOI 10.54499/UIDB/04020/2020 (https://doi.org/10.54499/UIDB/04020/2020).

Notes on contributors

Olugbenga Akinkunmi George

Olugbenga Akinkunmi George holds a Master in Tourism Organisation Management from the Faculty of Economics, Universidade do Algarve.

Célia M. Q. Ramos

Célia M. Q. Ramos holds a PhD in Quantitative Methods Applied to Economics and Management from Faculty of Economics, University of Algarve, obtained her Master in Electrical and Computers Engineering from the Higher Technical Institute, Lisbon University, and is graduated in Computer Engineering from the University of Coimbra, Portugal. She is Coordinator Professor at School for Management, Hospitality and Tourism, also in the UALG, where she lectures mainly Information Systems. Current research interests includes, information systems, data analytics, electronic tourism, business intelligence, digital marketing, panel data models. Researcher at the Centre for Tourism, Sustainability and Well-being (CinTurs). ORCID ID: 0000-0002-3413-4897.