Harnessing Social Media Data for Sentiment Analysis of Tourist
Attractions in Trat Province, Thailand using the Random Forest Machine
Learning Approach
NARONG PLEERUX1, PHANNIPHA ANURUKSAKORNKUL2,
PARADORN BOONPOR3, PARINYA NAKPATHOM4
1Faculty of Geoinfomatics, Burapha University,
Chon Buri, THAILAND
2Faculty of Humanities and Social Sciences, Burapha University,
Chon Buri, THAILAND
3Department of National Parks, Wildlife and Plant Conservation,
Ministry of Natural Resources and Environment,
Bangkok, THAILAND
4Internatioal Collage, Burapha University,
Chon Buri, THAILAND
Abstract: Tourism and service industries are vital economic drivers worldwide, and social media platforms play a pivotal role
in disseminating and gathering tourist reviews. This study employed the random forest algorithm to analyze tourist reviews of
attractions in Trat Province, Thailand, using data collected from the Tripadvisor website between 2014 and 2023. From the
results, key issues impacting these destinations were identified and categorized into four main areas, i.e., scenery, facilities,
safety, and accessibility. With a high accuracy rate of 99.65%, the analysis revealed that 98.66% of the reviews reflected
positive sentiment, underscoring the province’s appeal. However, the findings of this study also highlight critical challenges,
particularly in terms of facilities and safety, which require attention to realize sustainable tourism management. The findings
provide valuable insights for stakeholders to enhance the quality of tourism services in Trat, aligning with the province’s
aspirations to elevate its status to a primary tourist destination in Thailand.
Keywords: natural language processing, sentiment analysis, random forest, social media, tourism
Received: April 9, 2024. Revised: September 15, 2024. Accepted: November 16, 2024. Published: December 31, 2024.
1. Introduction
The tourism and service sectors are vital sources of
income for nearly all nations globally [1]. As tourism has
grown, there has been a corresponding increase in tourism-
related websites and social networks, which has resulted in a
substantial increase in the creation and sharing of tourism-
related information and opinions [2]. Many people express
their views and seek information via various social media
platforms [3], e.g., X (formerly Twitter), Facebook, and
Tripadvisor [4–5]. The global popularity of social media
platforms continues to grow, playing an increasingly integral
role in people’s lives across all dimensions [6], and such
platforms have become the primary mode of communication
for people worldwide [7].
Information shared on social media appears in various
formats, including text, images, and videos, with most
content reflecting opinions on events or topics, e.g., reviews
of tourist attractions on Tripadvisor [8–9] or photo sharing
on Instagram [10]. Social media platforms enable users to
express their opinions openly and directly. Social media data
offers real-time information and is cost-effective compared
to traditional survey methods [6]. In addition, social media
data can cover extensive time periods and allow for the
selection of specific time frames; thus, social media data are
ideal for studying the development or behavior of individuals
and groups over different periods [11].
Data from social media are utilized to examine people’s
satisfaction, attitudes, and emotions toward products,
services, or different issues through sentiment analysis. Such
analyses are typically categorized into three levels,. i.e.,
positive, negative, and neutral [12–13]. Sentiment analysis is
widely applicable in various fields, particularly in tourism
and service industries, and it is frequently used to analyze
customer opinions about restaurants [14–15], hotels [16–19],
and tourist destinations [20–21].
The primary objective of this study is to analyze tourist
feedback about attractions in Trat Province, Thailand, using
machine learning models and identify key issues affecting
these destinations. According to the Tourism Authority of
Thailand, Trat is classified as a secondary tourist province in
Thailand. A secondary tourist province is defined as one that
receives fewer than four million tourists per year [22]. The
Thai government has been making efforts to elevate these
provinces to primary tourist destinations by implementing
various promotional measures, e.g., tax incentives [23] and
organizing events to attract more visitors. Consequently,
tourist feedback from social media platforms like Tripadvisor
is a crucial source of information to manage and plan tourism
effectively in Trat Province according to current needs and
circumstances.
DESIGN, CONSTRUCTION, MAINTENANCE
DOI: 10.37394/232022.2024.4.23
Narong Pleerux, Phannipha Anuruksakornkul,
Paradorn Boonpor, Parinya Nakpathom
E-ISSN: 2732-9984
216
Volume 4, 2024
The remainder of this paper is organized as follows.
Section 2 presents the target data and methodology used in
this study. Section 3 discusses the research findings and
corresponding analysis. Finally, the paper is concluded in
Section 4.
2. Data and Methodology
2.1 Study Area
Trat Province, located at the easternmost tip of Thailand,
is shaped like an elephant’s head and spans 2,819 km2. It is
the fourth smallest province in the eastern region of the
country and ranks 56th in size nationwide. The province is
315 km from Bangkok.
Trat is known for its rich biodiversity, featuring
waterfalls, mountains, the sea, and beautiful coral reefs, as
well as abundant natural resources. Trat Province also has a
rich history and is home to internationally famous tourist
spots, e.g., Koh Chang and Koh Kood (Koh means island).
In addition, there are several notable community-based
tourism sites, e.g., Ban Nam Chiao and Ban Tha Ranae. Trat
is also an important area for fruit cultivation in Thailand.
2.2 Data Collection
In this study, tourist reviews of attractions in Trat
Province from 2014 to 2023 were gathered from Tripadvisor.
The data collection process was performed by web scraping
using Python with the Selenium and Beautiful Soup libraries
[24].
The acquired dataset comprised a total of 8,492
Tripadvisor reviews, each containing the name of the
attraction, the date of the review, and the review content.
Note that only reviews written in English were considered in
this study.
2.3 Data Preprocessing
After collecting the tourist reviews of attractions in Trat
Province, the data were preprocessed using natural language
processing techniques to clean the text. Here, the first step
involved converting all text to lowercase English. Then,
irrelevant characters, e.g., punctuation, URLs, symbols,
numbers, and special characters, were removed. The text was
then tokenized into smaller linguistic units. Finally, a list of
stop words was applied to filter out insignificant words,
including prepositions, conjunctions, pronouns, classifiers,
and emojis (e.g., “I,” “you,” “we,” “me,” “the,” and “is”).
2.4 Sentiment Analysis
The sentiment analysis in this study involves five
essential steps: review labeling, review splitting, which is
vital for preparing data for training and testing the machine
learning model, text representation, model development for
sentiment analysis, and model performance evaluation. The
specifics of these steps are outlined below.
1) The dataset of 8,492 tourist reviews of attractions in
Trat Province was used for the review labeling process.
Here, a random sample of 10% of the reviews (849 reviews)
was selected, and then three experts in tourism and data
science categorized the reviews into positive, neutral, and
negative sentiment groups. The labeled data were then
utilized to train a machine learning model.
2) Review splitting is an essential process to prepare
data to train and test a machine learning model. The review
dataset was divided into three subsets, i.e., a training set, a
validation set, and a test set with respective proportions of
64%, 16%, and 20%. The model utilized these data to learn
and identify patterns and relationships. The validation set
was used to evaluate and refine the model throughout the
training process, and the test set was used to assess the
model’s performance after training and validation.
Importantly, the test set is completely distinct from both the
training and validation sets, which ensures that the model
does not encounter any data from the test set.
3) The text representation process is a critical step in
preparing inherently nonnumerical text. Converting text into
numerical data is essential to realize effective analysis. This
study utilized the term frequency-inverse document
frequency vectorizer (TFIDFVectorizer) as the text
representation method. This technique considers both TF
and the significance of words across different documents
(i.e., the IDF) [25].
4) The random forest (RF) algorithm was employed for
sentiment analysis. Introduced by Breiman in 2001 [26], the
RF algorithm integrates the principles of random subspaces
and bagging. The decision tree forest algorithm is trained on
multiple decision trees, each of which uses slightly different
subsets of the data [27]. The RF method is adept at handling
complex and diverse datasets, and it effectively mitigates
overfitting problems.
The compound score for each review generated by the
model ranges from −1 to 1 and is categorized into three
groups, i.e., positive (≥0.05), neutral (≥−0.05 and <0.05),
and negative (<−0.05) [28].
5) The evaluation of the model’s performance includes a
comprehensive overview of its effectiveness, incorporating
several metrics, e.g., accuracy, precision, recall, and F1-
score [29].
In sentiment analysis model evaluation, accuracy,
precision, recall, and F1-score are crucial metrics for
measuring performance. While accuracy reflects the
proportion of correctly classified sentiments overall, it can be
misleading in imbalanced datasets. Precision highlights the
correctness of predicted positive sentiments, and recall
shows how well the model identifies actual positives. Since
these metrics can trade off, the F1-score combines them into
a single value, offering a more balanced measure, especially
when dealing with imbalanced data or when both false
positives and false negatives have different impacts.
Together, these metrics offer a well-rounded evaluation of
model performance in sentiment classification.
2.5 Identification of Issues
This section categorizes the issues identified at the target
tourist attractions in Trat Province based on the negative
review from tourists, which were analyzed using the RF
model. The issues were grouped into four categories, i.e.,
scenery (representing the beauty and cleanliness of the
attractions), facilities (the adequacy of amenities, staff, and
services), safety (crime and the safety of the services
provided at the attractions), and accessibility (representing
DESIGN, CONSTRUCTION, MAINTENANCE
DOI: 10.37394/232022.2024.4.23
Narong Pleerux, Phannipha Anuruksakornkul,
Paradorn Boonpor, Parinya Nakpathom
E-ISSN: 2732-9984
217
Volume 4, 2024
ease of use, affordability, and the availability of multiple
options to access attractions).
3. Results and Discussion
The findings are discussed in terms of three perspectives,
i.e., the number of tourist reviews, the results of analyzing
the tourist reviews of attractions using the RF model, and the
categorization of the identified issues at the attractions based
on negative reviews.
3.1 Number of Tourist Reviews
The analysis of tourist comments of attractions in Trat
Province, which were collected from the Tripadvisor website
from 2014 to 2023, revealed a total of 8,492 comments. The
peak year for comments was 2016, with 1,628 comments,
which was followed by 2017 with 1,321 comments and 2019
with 1,215 comments. However, during the COVID-19
pandemic, there was a considerable decline in the number of
comments, with only 133 recorded in 2021. This decrease
was largely due to the government’s lockdown measures and
international travel bans implemented to mitigate the spread
of the virus [30], thereby resulting in a significant drop in
travel and commentary. As the pandemic situation improved,
the number of comments began to increase steadily, as
shown in Fig. 1.
Fig. 1. Annual number of tourist reviews on Tripadvisor from 2014 to
2023.
An analysis of the top 10 attractions with the most
reviews revealed that six are natural attractions, and the
remaining four are diving agencies. For example, Klong Plu
Waterfall was found to be the most reviewed site, with 858
reviews (representing 10.10% of the total reviews).
Following this, White Sand Beach was reviewed 697 times
(representing 8.21% of the total reviews), and the Scuba
Dawgs diving school ranked third with 454 (5.36%). Note
that most of the top 10 reviewed attractions are natural
attractions. This can be attributed to Trat Province’s location
along Thailand’s eastern coast, which boasts stunning and
famous beaches and islands, e.g., Koh Chang, where Klong
Plu Waterfall, Khlong Prao Beach, and Bang Bao Beach are
located. Other attractions, e.g., Koh Kood and Koh Mak, also
draw significant tourist interest, resulting in numerous
reviews of these attractions. In addition, four of the top 10
sites are diving agencies, reflecting the province’s clear
waters, diverse marine life, and colorful coral reefs, which
make it a popular destination for snorkeling and scuba
diving. Both Thai and international tourists frequently visit
dive sites, e.g., Koh Rang, Koh Chang, and Blueberry Hill,
as shown in Table I.
TABLE I. TOP 10 ATTRACTIONS IN TRAT PROVINCE WITH THE
HIGHEST NUMBER OF TRIPADVISOR REVIEWS
Ranking
Attractions
Frequency
1st
Klong Plu Waterfall
858
2nd
White Sand Beach
697
3rd
Scuba Dawgs
455
4th
BB Divers
388
5th
Khlong Prao Beach
329
6th
BB Divers Koh Kood
297
7th
Koh Kood Divers
289
8th
Bang Bao Beach
246
9th
Lonely Beach
240
10th
Kai Bae Beach
238
a. Percentage of 8,492 reviews
3.2 Sentiment of Reviews
The examination of the tourist reviews of attractions in
Trat Province indicates that the model achieved an overall
accuracy of 99.65%. Remarkably, 98.66% of the reviews
(totaling 8,378 reviews) conveyed positive sentiments
toward the attractions in the region. In comparison, neutral
reviews accounted for 0.73% (62 reviews), and negative
review comprised only 52 reviews (representing 0.61%), as
shown in Table II.
TABLE II. NUMBER OF POSITIVE, NEUTRAL, AND NEGATIVE TOURIST
REVIEWS OF ATTRACTIONS IN TRAT PROVINCE OBTAINED BY THE RF
MODEL
Year
Positive
Neutral
Negative
Total
2014
623
6
2
631
2015
1178
9
10
1,197
2016
1596
22
10
1,628
2017
1313
3
5
1,321
2018
1153
8
4
1,165
2019
1189
10
16
1,215
2020
513
1
2
516
2021
131
0
2
133
2022
303
0
0
303
2023
379
3
1
383
Total
8,378
62
52
8,492
%
98.66
0.73
0.61
100.00
After performing a detailed analysis of the 8,492 reviews,
we found that they pertained to 189 distinct tourist attractions
and service establishments. Most tourists expressed positive
feedback about various aspects of these attractions. For
example, Klong Plu Waterfall was praised for its beauty and
easy of accessibility, and White Sand Beach received
comments highlighting its long stretch of white sandy shore
and clear, clean waters, thereby making it an ideal
DESIGN, CONSTRUCTION, MAINTENANCE
DOI: 10.37394/232022.2024.4.23
Narong Pleerux, Phannipha Anuruksakornkul,
Paradorn Boonpor, Parinya Nakpathom
E-ISSN: 2732-9984
218
Volume 4, 2024
destination for families with children. In addition, some
tourists noted that White Sand Beach has numerous street
food vendors and excellent restaurants.
In contrast, several issues were identified when analyzing
the negative feedback about these attractions, including
cleanliness problems, service quality concerns, and incidents
of animal cruelty. Specific issues at White Sand Beach
included problems related to the beach’s condition and
cleanliness, safety hazards, and the availability, quality, and
ethical standards of accommodations and services. For
example, at elephant camps, tourists reported an
unwelcoming atmosphere, poor service, and concerns about
animal abuse and ethical practices. Even Klong Plu
Waterfall, which received the most positive feedback, as
mentioned previously, was not without its issues. The
problems cited included management inefficiencies, cost and
value concerns, and safety issues, particularly regarding the
safety of children.
3.3 Issues Identification from Negative Reviews
To gather information about the identified issues,
complaints, and suggestions from tourists regarding the
attractions in Trat Province, we analyzed and categorized 52
negative reviews into four main areas, i.e., scenery, facilities,
safety, and accessibility. Note that a single comment may
address multiple issues; thus, we counted the total number of
distinct issues described in each comment.
Among the negative reviews, the identified issues were
related to 25 different attractions and services, with a total of
93 mentions across the four identified categories. The most
frequently mentioned issue was facilities, which was
referenced 37 times, accounting for 39.78% of the total. This
was followed by safety, which was mentioned 26 times
(27.96%), and scenery, which appeared 19 times (20.43%).
We found that accessibility was the least discussed issue,
with only 11 mentions (representing 11.83%), as shown in
Table III.
TABLE III. NUMBER AND PERCENTAGE OF ISSUES IDENTIFIED AT
TOURIST ATTRACTIONS IN TRAT PROVINCE DERIVED FROM NEGATIVE
REVIEWS ANALYZED USING THE RF MODEL
Issues
Natural
attractions
Man-made
attractions
Total
No.
%
No.
%
No.
%
Scenery
12
32.43
7
12.51
19
20.43
Facilities
10
27.03
27
48.21
37
39.78
Safety
8
21.62
18
32.14
26
27.96
Accessibility
7
18.92
4
7.14
11
11.83
Total
37
100.00
56
100.00
93
100.00
The 25 tourist attractions mentioned previously can be
classified into two main categories, i.e., natural attractions
and man-made attractions, which helps facilitate the
discussion of various issues. Focusing on natural attractions,
tourists provided 20 negative reviews (out of a total of 52)
that were associated with seven different natural sites.
Among these sites, White Sand Beach received the highest
number of negative comments, with seven in total. This was
followed by Klong Plu Waterfall and Lonely Beach, each of
which received four negative reviews. The remaining sites,
i.e., Bang Bao Beach, Wai Chaek Beach, Than Mayom
Waterfall, Ao Noi Beach, and Kai Bae Beach, were each
mentioned only once in a negative context.
Among the negative reviews related to natural attractions,
tourists most frequently mentioned concerns about scenery,
accounting for 32.43% of the feedback. This was followed
by issues regarding facilities (27.03%), safety (21.62%), and
accessibility (18.92%). Regarding scenery, tourists primarily
criticized the cleanliness, especially at several beaches, e.g.,
White Sand Beach, Lonely Beach, and Kai Bae Beach,
where problems with litter and trash were identified.
In terms of facilities, tourists expressed dissatisfaction
with the entrance fees at Klong Plu Waterfall and Than
Mayom Waterfall, which were considered too high (200 and
100 baht or approximately 5.40 and 2.70 US dollar for
foreign adults and children, respectively). Note that these
waterfalls are located within a national park, and the fees are
set according to the park’s regulations.
Concerning safety and accessibility, tourists were
particularly concerned about the steep roads on Koh Chang,
given the island’s hilly terrain interspersed with flat areas,
which could contribute to accidents. In addition, concerns
were raised about the overall quality of the roads.
For man-made attractions, tourists provided 32 negative
reviews out of a total of 52. The most frequently criticized
aspect was the facilities, accounting for 48.21% of the
negative feedback, followed by safety concerns at 32.14%,
and scenery at 12.51%. Accessibility received the fewest
negative mentions, at only 7.14%. When analyzing the issues
at these man-made attractions, the majority were found to be
related to tour and diving agencies, restaurants, pubs and
bars, and animal camps.
Regarding issues with facilities, many tourists reported
negative experiences, e.g., poor service and a lack of
responsibility from staff at diving agencies. In restaurants,
pubs, and bars, complaints included rude and unprofessional
behavior from staff, and poor cleanliness and sanitation on
buses and boats. Another significant concern was related to
animal camps, particularly elephant camps, which are
prevalent in Trat Province. Tourists expressed ethical
concerns about riding elephants, the abuse and distress of the
animals, and their overall poor treatment.
The next most significant issue at man-made attractions
and services was safety, with several key concerns identified,
including food hygiene and safety in restaurants, as well as
dangerous diving practices and instructor attitudes impacting
safety at tour and diving agencies. In terms of scenery,
tourists mentioned an unwelcoming atmosphere and dirty
locations. The final issue, i.e., accessibility, was primarily
focused on ferry services, with complaints about
overcrowded ferries and chaotic transfers.
3.4 Theoretical Implications
This study employed the RF algorithm to analyze tourist
reviews of attractions in Trat Province using review data
acquired from the Tripadvisor website. The RF algorithm is
particularly advantageous when handling large volumes of
unstructured textual reviews [31]. It combines high accuracy
with relatively quick training times; thus, the RF algorithm is
ideal for complex sentiment analysis that requires both
precision and efficiency [32]. The RF algorithm can manage
both regression and classification tasks with a high degree of
DESIGN, CONSTRUCTION, MAINTENANCE
DOI: 10.37394/232022.2024.4.23
Narong Pleerux, Phannipha Anuruksakornkul,
Paradorn Boonpor, Parinya Nakpathom
E-ISSN: 2732-9984
219
Volume 4, 2024
accuracy and a reduced risk of overfitting [33]. Its key
advantage lies in the high accuracy of its results, which has
resulted in its widespread use in sentiment analysis, as
exhibited by the following accuracy rates: 99.04% [32],
82.91% [34], 86.00% [35], and 83.50% [36]. In the current
study, the RF algorithm achieved an accuracy of 99.65%,
further underscoring its effectiveness in terms of applying
machine learning technology to analyze sentiment from
social media, particularly in the context of tourism and
service-related feedback.
3.5 Managerial Implications
This study performed an in-depth analysis of the negative
tourist feedback, categorizing it into specific issues to
identify problems faced at tourist attractions in Trat
Province, as directly reported by the visitors. The insights
gained from this analysis can inform strategies to address and
managing these challenges. For example, the most frequently
mentioned issue involved facilities, with complaints focusing
on negative service experiences, e.g., impolite and
unaccommodating staff, as well as cleanliness concerns in
restrooms, boats, and buses. Safety was the second most
significant concern, divided into two key areas, i.e., food
hygiene and the safety of services, particularly in activities
like diving. To address these concerns, businesses, e.g.,
restaurants, ferry operators, and tour and diving companies,
should establish, implement, and maintain clear service
standards and provide staff effective training on both
customer etiquette and diving safety.
Issues related to the natural scenery were primarily
observed at nature-based attractions, where the main
problems were related to cleanliness and litter on the
beaches. The litter was traced back to two major sources, i.e.,
marine debris, particularly during the monsoon season when
large amounts of trash are washed ashore, and waste left by
both tourists and locals. To address this, relevant authorities,
e.g., the local municipal government, should develop and
implement comprehensive plans to manage cleanliness,
including providing adequate trash bins and increasing the
frequency of beach cleanups.
Effectively and sustainably resolving these four key
issues requires collaborative efforts from all stakeholders,
including tourism business operators, government agencies,
local residents, and tourists. Only through such cooperation
can these challenges be addressed successfully.
4. Conclusion
In this study, the sentiment analysis technique was used
to evaluate the satisfaction and emotions of tourists or
customers regarding tourism and services in Trat Province,
Thailand. This study employed the RF algorithm, enhanced
by machine learning, to analyze tourist reviews of attractions
using data acquired from the Tripadvisor website. The results
demonstrate that a significant majority of tourists,
approximately 98.66%, expressed positive views about the
attractions in the target region. This high level of positive
sentiment indicates that the attractions are both beautiful and
appealing, successfully captivating and impressing tourists.
Nevertheless, this positive feedback can also guide further
improvements and enhancements to the province’s tourism
offerings. Conversely, the negative feedback can be
categorized into several key issues, with facilities being the
top priority, followed by safety, scenery, and accessibility
issues. These insights, including comments, criticisms, and
suggestions, can be relayed to relevant authorities to aid in
managing, planning, and addressing the challenges at tourist
attractions in Trat Province to better fulfill the needs of
tourists.
This study has two key limitations. First, the RF
algorithm was employed to analyze social media feedback,
and highly accurate results were obtained; however, future
research should consider incorporating additional models,
e.g., naïve Bayes, support vector machine, gradient boosting,
and other models, to optimize the outcomes. Second, this
study relied exclusively on reviews from the Tripadvisor
website, which may lead to potential bias. Therefore, to
achieve more comprehensive and balanced results, future
research should include data from other social media
platforms, e.g., X (formerly Twitter), and Facebook.
Acknowledgment
This research project was financed by the Fundamental
Fund 2024, Burapha University.
References
[1] S. Ainin, A. Feizollaha, N.B. Anuarb, and N.A. Abdullah, “Sentiment
analyses of multilingual tweets on halal tourism,” Tour. Manag.
Perspect., vol. 34, pp. 100658, April 2020.
[2] M. Ghane’i‑Ostad, H. Vahdat‑Nejad, and M.
Abdolrazzagh‑Nezhad, “Detecting overlapping communities in
LBSNs by fuzzy subtractive clustering,” Soc. Netw. Anal. Min., vol.
8, pp. 23, March 2018.
[3] D. Arora, K.F. Li, and S.W. Neville, “Consumers' sentiment analysis
of popular phone brands and operating system preference using
twitter data: A feasibility study,” The 2015 IEEE 29th International
Conference on Advanced Information Networking and Applications,
pp. 680–686, March 2015.
[4] N. Leelawat, S. Jariyapongpaiboon, A. Promjun, S. Boonyarak, K.
Saengtabtim, A. Laosunthara, A.K. Yudha, and J. Tang, “Twitter data
sentiment analysis of tourism in Thailand during the COVID-19
pandemic using machine learning,” Heliyon, vol. 8, pp. e10894,
October 2022.
[5] E. Polyzos, A. Fotiadis, and T.C. Huan, “The asymmetric impact of
Twitter sentiment and emotions: Impulse response analysis on
European tourism firms using micro-data,” Tour. Manag., vol. 104,
pp. 104909, October 2024.
[6] Y. Cui, and Q. He, Travel behavior analysis with social media data
and smartphone GPS data. Tech Report, CA: University of Buffao,
2019.
[7] N.B. Lassen, R. Madsen, and R. Vatrapu, “Predicting iphone sales
from iphone tweets,” The 2014 IEEE 18th International Enterprise
Distributed Object Computing Conference, pp. 81–90, September
2014.
[8] A. Valdivia, M.V. Luzon, and F. Herrera, “Sentiment analysis in
TripAdvisor,” IEEE Intell. Syst., vol. 32, pp. 72–77, August 2017.
[9] N. Pleerux, and A. Nardkulpat, “Sentiment analysis of restaurant
customer satisfaction during COVID-19 pandemic in Pattaya,
Thailand,” Heliyon, vol. 9, pp. e22193, November 2023.
[10] T. Bastrygina, W.M. Lim, R. Jopp, and M.A. Weissmann,
“Unraveling the power of social media influencers: Qualitative
insights into the role of Instagram influencers in the hospitality and
tourism industry,” J. Hosp. Tour. Manag., vol 58, pp. 214–243, March
2024.
[11] M. Picornell, T. Ruiz, M. Lenormand, J.J. Ramasco, T. Dubernet, and
E. Frias-Martines, “Exploring the potential of phone call data to
characterize the relationship between social network and travel
behavior,” Transportation, vol. 42, pp. 647–668, July 2015.
[12] A. Valdivia, M.V. Luzon, and F. Herrera, “Sentiment analysis on
TripAdvisor: are there inconsistencies in user reviews?,” “The
DESIGN, CONSTRUCTION, MAINTENANCE
DOI: 10.37394/232022.2024.4.23
Narong Pleerux, Phannipha Anuruksakornkul,
Paradorn Boonpor, Parinya Nakpathom
E-ISSN: 2732-9984
220
Volume 4, 2024
12th International Conference on Hybrid Artificial Intelligent
Systems,” pp. 15–25, June 2017.
[13] D. Obembe, O. Kolade, F. Obembe, A. Owoseni, and O. Mafimisebi,
“Covid-19 and the tourism industry: An early stage sentiment analysis
of the impact of social media and stakeholder communication,” Int. J.
Inf. Manag. Data Insights, vol. 1, pp. 100040, November 2021.
[14] H. Zhang, P. Ji, J. Wang, and X. Chen, “A novel decision support
model for satisfactory restaurants utilizing social information: a case
study of TripAdvisor.com,” Tour. Manag., vol. 59, pp. 281–297,
April 2017.
[15] P. Pleerux, W. Boonpook. A. Chaiyo, and A. Nardkulpat,
“TripAdvisor restaurant customer reviews analysis in Bangkok,
Thailand,” The 2023 International Conference on Cyber Management
and Engineering, pp. 536–539, January 2023.
[16] C. Tsai, K. Chen, Y. Hu, and W. Chen, “Improving text
summarization of online hotel reviews with review helpfulness and
sentiment,” Tour. Manag., vol. 80, pp. 104122, October 2020.
[17] Y. Chang, C. Ku, and C. Chen, “Using deep learning and visual
analytics to explore hotel reviews and responses,” Tour. Manag., vol.
80, pp. 104129, October 2020.
[18] J. Zhang, X. Lu, and D. Liu, “Deriving customer preferences for
hotels based on aspect-level sentiment analysis of online reviews,”
Electron. Commer. Res. Appl., vol. 49, pp. 101094, September–
October 2021.
[19] Y. Bian, R. Ye, J. Zhang, and X. Yan, “Customer preference
identification from hotel online reviews: A neural network based fine-
grained sentiment analysis,” Comput. Ind. Eng., vol. 172, pp. 108648,
October 2022.
[20] N.G. Uğur, and A. Akbıyık, “Impacts of COVID-19 on global
tourism industry: A cross-regional comparison,” Tour. Manag.
Perspect., vol. 36, pp. 100744, October 2020.
[21] Z. Abbasi-Moud, H. Vahdat-Nejad, and J. Sadri, “Tourism
recommendation system based on semantic clustering and sentiment
analysis,” Expert Syst. Appl., vol. 167, pp. 114324, April 2021.
[22] TAT Intelligence Center, (2018). “The compass of Tourism 4/2018,”
[online] Available at:<https://www.dla.go.th/upload/news/type9/2018/
11/40628_1.pdf ?time=1544920032593>
[23] Ratchakitcha, (2019). “Ministerial Regulation No. 344 B.E. 2562
(2019) issued under the Revenue Code regarding tax exemption
Volume 136 Part 65 A,” [Online] Available at: <https://www.rd.go.th/
fileadmin/user_upload/kormor/newlaw/mr344.pdf>
[24] S. Han, and C.K. Anderson, “Web Scraping for Hospitality Research:
Overview, Opportunities, and Implications,” Cornell Hosp. Q., vol.
62, pp. 89–104, July 2020.
[25] M. Arora, V. Mittal, and P. Aggarwal, “Enactment of tf-idf and
word2vec on text categorization,” The 3rd International Conference
on Computing Informatics and Networks, pp. 199–209, March 2019.
[26] L. Breiman, “Random Forests,” Mach. Learn., vol. 45, pp. 5–32,
October 2001.
[27] Y. Al Amrani, M. Lazaar, and K.E. El Kariri, “Random forest and
support vector machine based hybrid approach to sentiment analysis,”
Procedia Comput. Sci., vol. 127, pp. 511–520, 2018.
[28] A.R. Pai, M. Prince, and C.V. Prasannakumar, “Real-time Twitter
sentiment analytics and visualization using Vader”, The 2nd
International Conference on Intelligent Technologies, pp. 1–4, June
2022.
[29] O. Rainio, J. Teuho and R. Klén, “Evaluation metrics and statistical
tests for machine learning,” Sci. Rep., vol. 14, pp. 6086, July 2024.
[30] Ratchakitcha, (2021). “Regulation issued under section 9 of the
emergency decree on public administration in emergency situations
B.E. 2548 (2005) (No. 1),” [Online] Available at: <http://www.ratcha
kitcha.soc.go.th/DATA/PDF/2563/E/ 069/T_0010.PDF>
[31] N. Saraswathi, T.S. Rooba, and S. Chakaravarthi, “Improving the
accuracy of sentiment analysis using a linguistic rule-based feature
selection method in tourism reviews,” Meas.: Sens., vol. 29, pp.
100888, October 2023.
[32] T. Anderson, S. Sarkar, and R. Kelley, “Analyzing public sentiment
on sustainability: A comprehensive review and application of
sentiment analysis techniques,” J. Nat. Lang. Process., vol. 8, pp.
100097, September 2024.
[33] IBM, (2024). “What is random forest?,” [Online] Available at:
<https://www.ibm.com/topics/random-forest>.
[34] V.A. Fitri, R. Andreswari, M.A. Hasibuan, “Sentiment analysis of
social media Twitter with case of anti-LGBT campaign in Indonesia
using Naïve Bayes, Decision Tree, and Random Forest algorithm,”
Procedia Comput. Sci., vol. 161, pp.765–772, July 2019.
[35] M. Shahzad, C. Freeman, M. Rahimi and H. Alhoori, “Predicting
Facebook sentiments towards research,” J. Nat. Lang. Process., vol. 3,
pp. 100010, June 2023.
[36] M. Ramzy & B. Ibrahim, “User satisfaction with Arabic COVID-19
apps: Sentiment analysis of users’ reviews using machine learning
techniques,” Inf. Process. Manage., vol. 61, pp. 103644, May 2024.
DESIGN, CONSTRUCTION, MAINTENANCE
DOI: 10.37394/232022.2024.4.23
Narong Pleerux, Phannipha Anuruksakornkul,
Paradorn Boonpor, Parinya Nakpathom
E-ISSN: 2732-9984
221
Volume 4, 2024
Contribution of Individual Authors to the
Creation of a Scientific Article (Ghostwriting
Policy)
The authors equally contributed in the present
research, at all stages from the formulation of the
problem to the final findings and solution.
Sources of Funding for Research Presented in a
Scientific Article or Scientific Article Itself
No funding was received for conducting this study.
Conflict of Interest
The authors have no conflicts of interest to declare
that are relevant to the content of this article.
Creative Commons Attribution License 4.0
(Attribution 4.0 International, CC BY 4.0)
This article is published under the terms of the
Creative Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.en
_US