A sentiment analysis model of agritech startup on Facebook comments using naive Bayes classifier

International Journal of Electrical and Computer Engineering (IJECE)
Vol. 12, No. 3, June 2022, pp. 2829~2838
ISSN: 2088-8708, DOI: 10.11591/ijece.v12i3.pp2829-2838  2829
Journal homepage: http://ijece.iaescore.com
A sentiment analysis model of agritech startup on Facebook
comments using naive Bayes classifier
Nawapon Kewsuwun, Siriwan Kajornkasirat
Faculty of Science and Industrial Technology, Prince of Songkla University, Surat Thani, Thailand
Article Info ABSTRACT
Article history:
Received May 16, 2021
Revised Dec 31, 2021
Accepted Jan 19, 2022
Facebook page is a tool able to generate perceptions and acceptance, and
support people and investors in making business decisions. Moreover,
Facebook page plays a part in engaging people in the form of a community.
People share experiences and opinions toward products, services, and trends
in particular periods on the Facebook page community. Regarding sentiment
analysis on Facebook pages, most education and other general topics in
English have only been analyzed in English. However, sentiment analysis
regarding agritech startups topics in Thai language has not been done yet.
This study analyzes opinions and categorizes positive and negative
comments by using naive Bayes classifier to examine the sentiments and
attitudes of people and investors. The results could possibly reflect the
perception rate of agritech startups in Thailand and could be applied to
explain attentiveness and assess people’s engagement opinions.
Furthermore, it could be applied in studying consumer behavior, marketing
analysis, spread of information, and attitudes. The study's model is generic
and could be applied in other contexts to provide insightful suggestions.
Keywords:
Agritech startup
Facebook
Naive Bayes
Python
Sentiment analysis
This is an open access article under the CC BY-SA license.
Corresponding Author:
Siriwan Kajornkasirat
Faculty of Science and Industrial Technology, Prince of Songkla University
31 Moo 6, Makhamtia, Muang, Surat Thani 84000, Thailand
Email: siriwan.wo@psu.ac.th
1. INTRODUCTION
Digital information is consistently accessed at increasing rates [1]. Facebook is one of the digital
platforms where people post information relating to behaviors, attitudes, and feelings. Comments on social
media platforms might be useful since they may reflect social concerns, norms, trends, or opportunities in
accepting or objecting [2], [3]. Therefore, there is a good opportunity to analyze comments from Facebook to
gain helpful information regarding a particular context, such as the agritech startups in Thailand.
Sentiment analysis studies and analyzes people's responses and acceptance toward an entity (e.g.,
posts, blogs, comments, book reviews, videos) using text analysis by computational algorithms to help
determine textual reactions whether they are positive, negative, or neutral [4]. Moreover, sentiment analysis
can analyze or predict trends, for example, in the stock market [5], election polls [6], or in other contexts [7].
Notably, people make comments, give opinions or share their experiences regarding service and products
with friends and acquaintances through an online community. Consequently, educational and business sector
are interested in studying customer opinions on online platforms. Nevertheless, the surveying might take time
in collecting data, and opinions might change [8].
Researchers in computer technology and smart business have raised the aforementioned concerns
[8], [9]. They designed an easy and fast algorithm to collect and analyze opinions; further, it is reliable and
valid in terms of processing natural language and using machine learning to gather and assess the data [4],

 ISSN: 2088-8708
Int J Elec & Comp Eng, Vol. 12, No. 3, June 2022: 2829-2838
2830
[10]. Word analysis is necessary and interesting in this sentiment analysis research, especially for Thai
language, since it is a complex and unique language for its structures, word formations, and varieties.
Facebook is a well-known social media platform in Thailand. In 2020, it was found that Thailand
was ranked 8th
in spending time on Facebook, with people posting opinions and engaging in activities [11].
Facebook users include, for example, individuals, corporates, and communities from government, private
sector, and civil society. The Facebook posts include political news, entertainment news, and economic news
and are open for public comments. The comments are crucial in sentiment analysis because they reflect
people's thoughts toward those posts.
This study focused on analyzing and classifying positive and negative comments using a Naive
Bayes classifier for validity. The algorithm is fast and accurate when classifying large and complex data. Its
main attribute is being condition-free, as it only requires simple word parameter training. However, it is
difficult to identify and interpret the level of sentiment regarding neutral word conditions so that neutral
opinions were not counted. Thus, the researchers employ this method to classify positive and negative
opinions on Facebook regarding agritech startups of Thailand. Many studies shed light on sentiment analysis
in Thai language on Twitter, Facebook, and other platforms. For instance, a study of sentiment analysis using
deep learning techniques on Thai Twitter data [12], using sentiment analysis to analyze Thai customer
satisfaction from social media [13], and sentiment analysis of Thai children's stories [14] have used Naive
Bayes to explore sentiment issues.
This study analyzed comments regarding agritech startups in Thailand on Facebook pages from
April 2016 to March 2021 by gathering primary posts from the specific period. Agritech startups in Thailand
are one of Thailand's 4.0 policies, anticipating exponential improvement [15]. Therefore, the Facebook page
is a tool for building up perceptions, acceptance, and a support systems for people and investors [1].
Additionally, the Facebook page could engage and network people in a group format to share experiences
and opinions effectively, including metadata tags or hashtags [1]. Naive Bayes learning classifier is employed
to determine the percentages of words expressing emotions or positive and negative feelings.
The first part of this study presents sentiment analysis principles and procedures, and the second part
addresses agritech startups in Thailand. Finally, the last part reveals the findings of this study that might
reflect trends, views toward policies, support, or news on agritech startups in Thailand. More importantly, the
findings could identify future trends in developing agritech startups and interesting issues for investors and
the government.
2. LITERATURE REVIEW
Sentiment analysis focuses on examining and evaluating positivity and negativity of public
opinions, attitudes, and emotions toward a particular issue, for instance, services, products, events, or
situations [16]. Furthermore, the algorithm can classify and categorize positivity and negativity, which helps
make decisions [17]. Therefore, it reflects useful trends regarding attitudes and emotions of potential
customers. The study employs sentiment analysis using a naive Bayes classifier for facebook comments on
agritech startups in Thai language. The principles and concepts are as follows.
2.1. Sentiment analysis
Nowadays, sentiment analysis is a popular research topic in data science, computer science, and
business intelligence. Many real-life and data overload problems are part of this subject matter. It is also
highly challenging as a natural language processing (NLP) research topic covering many novel subproblems.
Additionally, there was almost no relevant research for sentiment analysis done before 2,000 on either NLP
or linguistics. This is because of the lack of availability of opinion or text in digital forms. Since the year
2,000, the field has multiplied to become one of the most active research areas in NLP [18]. It is also widely
researched in data mining, recommendation systems, expert systems, and information retrieval technologies
[2]. It has spread from data science to business intelligence [19]. There are mainly three methods or levels of
research on sentiment analysis [2].
– Document-level [20]: analyze the overall sentiment expressed in the text and determine if the overall
sentiment is positive or negative.
– Sentence level [21]: examine the sentiment expressed in sentences and determine whether each sentence
expresses a positive, negative, or neutral opinion.
– Aspect level [22]: aspect level performs better when document level and sentence level analyses do not
find what people liked and did not like. The aspect level is also called the feature level.
The aspect level does not work on document, paragraphs, sentences, or clauses; it directly finds the
target's sentiment. The sentiment can be positive, negative, or neutral towards that specific target or entity.

Int J Elec & Comp Eng ISSN: 2088-8708 
A sentiment analysis model of agritech startup on Facebook comments … (Nawapon Kewsuwun)
2831
Realizing the importance of opinion, targets also help us understand the sentiment analysis problem better.
Another challenge is to classify opinion into two types: regular opinion and comparative opinion [19].
The sentiment analysis studies presented in the literature are embedded in several topic areas:
i) Business e.g., brand reputation [23], product/service evaluation [18], consumers/patients voice [24];
ii) Politics, e.g., identifying voter mood during election [25]; and iii) Other, e.g., event monitoring, film
reviews, fake (spam) reviews [26].
2.2. Sentiment analysis on Facebook
In recent times, people can understand the real-world movements by analyzing social media data
(Facebook) [20]. Sentiment analysis tools have been applied to examine the relationship between the release
of products, the 'discussion' online, and comments to a post or on products, with the outcome being that such
data can be used to assess/predict the user opinions [21]. The sentiment analyses are about opinion or
expression or imply positive or negative sentiments or emotions towards an object. So, the elements of
sentiment analysis are [2], [27]: i) "Opinion" is a judgment, viewpoint, or statement about matters commonly
considered subjective [28]; ii) "entity," which is news, information, topic, issue, person, organization, or
event, is the target of an opinion; and iii) "subjectivity and emotions," which are the state of mind of a person
and instinctive responses.
2.3. Sentiment analysis on Facebook comments in Thai language
Facebook is one of the popular social networks in Thailand, on which people share opinions,
feelings, or knowledge. Therefore, Facebook status is an essential set of information in analyzing opinions
and feelings. To analyze posts in Thai language, naive Bayes was employed to pool comments and classify
positive or negative meaning [29]. Moreover, naive Bayes was employed to analyze reliability in facebook
status. The purpose of using naive Bayes is to identify attributes between status and relation of behavior and
specific feelings. However, analyzing sentiments in Thai language is complex and complicated at the
language level, meaning interpretation, and the use of the language [30]. Thus, the researchers considered
using naïve Bayes to ease the analysis.
2.4. Sentiment analysis using natural language processing
Sentiment analysis is natural language processing used to identify and analyze texts for whether the
sentiment is positive, negative, or neutral. The analysis helps businesses or investors examine and gain a
deeper understanding of customers' opinions and satisfaction, market trends, or economy [30]. Furthermore,
sentiment analysis using natural language processing and computing algorithms can simulate the sentiment
analysis models [31]:
− Rule-based: the system automatically performs sentiment analysis based on a set of manually crafted
rules. These rules may include various NLP techniques developed in computational linguistics: stemming,
tokenization, part-of-speech tagging and parsing, and lexicons (i.e., lists of words and expressions).
− Automatic methods: contrary to rule-based systems, these do not rely on manually crafted rules but on
machine learning techniques. A sentiment analysis task is usually modeled as a classification problem,
whereby a classifier is fed a text and returns a category, e.g., positive, negative, or neutral.
− Hybrid systems: combine the desirable elements of rule-based and automatic techniques into one system.
One huge benefit of these systems is that the results are often more accurate.
2.5. Sentiment analysis using Naive Bayes classifier
Many studies with machine learning algorithms have used sentiment or opinion mining. Some
common algorithms used in sentiment analysis are naïve bayes, maximum entropy, and support vector
machine [2], [29]. Naive Bayes classifiers are highly scalable and require a number of parameters linear in
the number of variables (features/predictors) in a learning problem. Maximum-likelihood training can be
done by evaluating a closed-form expression, which takes linear time rather than an expensive iterative
approximation used with many other classifiers [29], as in the equation:
𝑃 (𝐴|𝐵) =
𝑃 (𝐵|𝐴) ∗ 𝑃 (𝐴)
𝑃 (𝐵)
It was found that using a naive Bayes classifier is convenient and easy to do for word training,
especially for word features with large data sets to analyze and classify. In addition, it can be used to develop
complicated models. However, the naive Bayes classifier only performs well when the testing attributes are
independent [32]. Therefore, the researchers decided to employ this technique, because it can classify large
data sets, learn with a large number of inputs, and analyze the data in various ways. Thus, because the

 ISSN: 2088-8708
2832
collected large data sets are complicated and contain independent attributes, the researchers decided to use
naive Bayes classifier as the primary approach in this study.
2.6. Agriculture startups in Thailand
The startups and enterprises are important contributors in the national economy, the country's
lifeblood, and flexibly adapt to the circumstances; these are primary mechanisms in an economic recovery
and strengthen economic progress with increasing employment in society [33]. Moreover, agriculture startups
have become the training centers for workers who can gain real work experiences [34]. In addition, they are
starting points of product innovation and various types of services. People who are interested in investing in
startups and enterprises prefer lower capital and risk than in investing in large industries [33], due to
connecting networks and joint investments with large corporations or other related businesses. Additionally,
they create value for raw materials because they use domestic resources mainly generating income for the
country, directly in the agriculture, in manufacturing for export, and in the tourism sector [35]. Startups also
prevent monopoly formation in the economy, strengthen fair competition, and force efficiency of the overall
economy. More importantly, it would be beneficial to identify future trends in developing agritech startups
and interesting issues for investors and the government.
3. RESEARCH METHOD
This research applies a naive Bayes classifier to analysis of Facebook comments on agri-tech
startups, in Thailand. The researchers setting the process steps are: finding data, data collection, data
selection, data cleaning, data classification, and presentation of output. The research scheme is summarized
in Figure 1.
Figure 1. Research method
a. Finding data: the researchers searched for posts related to the agritech startups, in Thai language on
Facebook, using these keywords: agritech startups, startups, and startup business. The result was that
there were 12 posts from April 2016 to March 2021.
b. Data collection: the researchers used export comments software to gather the comments of the 12 posts.
There were 1,001 comments in total, classified as in Table 1.
c. Data selection: the researchers selected relevant data by skimming.
d. Data cleaning: the researchers eliminated redundant, irrelevant, and incomplete comments, including
tagging, pictures, and emoji icons. After cleaning, there were 405 comments as shown in Table 2.
e. Data classification: the researchers used the model training of Naive Bayes classifier algorithm, a binary
sentiment analysis in Python from Abromberg [36], to analyze and summarize the comments in Thai
language. The process is schematically shown in Figures 2 and 3.
Table 1. Comment classification
Post Name/Topic Comment Count
The prime minister and faculty visited the digital economy promotion agency 159
All Asian countries go to agritech startups 27
Startup greenhouse 132
Only 10k for a smart farm!! 97
Traditional farmers might work harder 30
StartUp! Use leaves to make containers 34
Listenfield: algorithm helps to form a new agriculture 60
Digital startup exhibition (digital Thailand big bang: digital station) 286
Startup, homegrown vegetable in container carriers 86
Containers from leaves 11
Soil scanner 29
Chula startup, Covid-19 vaccine from protein in tobacco leaves 50

2833
Table 2. Data after cleaning
Post Name/Topic Comment Count
The prime minister and faculty visited the digital economy promotion agency 61
All asian countries go to agritech startups 19
Startup greenhouse 43
Only 10k for a smart farm!! 41
Traditional farmers might work harder 16
StartUp! Use leaves to make containers 27
Listenfield: algorithm helps to form a new agriculture 26
Digital startup exhibition (digital Thailand big bang: digital station) 84
Startup, homegrown vegetable in container carriers 40
Containers from leaves 9
Soil scanner 17
Chula startup, Covid-19 vaccine from protein in tobacco leaves 22
f. Presentation of output: the outputs are
1) Showing positive and negative comments as integer counts and percentages: i) pos=xxxx, ii) neg=xxxx,
iii) pos=xx.xx %, and iv) neg=xx.xx %
2) Showing positive and negative comments from keyword search: i) text: xxxx, ii) test_sent: xxxx, and iii)
tag: xxxx (pos or neg)
Figure 2. Illustrates the use of the algorithm to classify Facebook comments on agritech startups in Thailand
Figure 3. Sentiment analysis process

 ISSN: 2088-8708
2834
4. RESULTS AND DISCUSSION
According to the sentiment analysis findings, most comments showed positive results related to
Thailand 4.0 [15] and policy of supporting Thai startups to strengthen the business sector's competitive level
[34]. Most sentiment analysis results reveal positive trends that identify the interests and acceptance of
people and investors regarding agritech startups. Thus, overall, the results show people's interests in agritech
startups in Thailand and can be used to study further consumer behavior, marketing analysis, product designs,
trends, and attitudes. To classify Thai language comments in positive and negative statements, the
researchers did classify comments and comments separated by post topic.
4.1. Classify comments
The 405 comments were classified as follows. Showing positive and negative comments as integer
count and percentage, shown in Figure 4, pos=258 (63.7%) and neg=147 (36.3%). In the analyzed data, there
were 258 positive comments and 147 negative comments. In other words, the commenting people had more
positive emotions (63.7%) than negative emotions (36.3%) towards the issue as shown in Figure 5 and
Figure 6. Showing positive and negative comments from keyword search as shown in Table 3.
Figure 4. Code sample to classify positive and negative comments as integer count and percentage
Figure 5. A sample of classifying positive and negative comments

2835
Figure 6. Code sample of classifying positive and negative comments in keyword search format
Table 3. Overall sentiment analysis
Comments Source
Comment Count Pos Sentiment Neg Sentiment
405
(100%)
258
(63.7%)
147
(36.3%)
According to the output, the researchers determined F-measure, precision, recall, and accuracy listed
in Table 4. Table 4 shows the effectiveness of the method in gathering and analyzing data. In this case study,
F-measure was 88.00%, precision was 85.00%, recall was 92.00%, and accuracy was 79.00%. Overall, the
researchers found that F-measure was 75.00%, precision was 80.00%, recall was 75.00%, and accuracy was
61.00%. This shows that some comments had slang or ill-structured phrases, which affected the word
analysis algorithm [37].
Table 4. Classifying comments by F-measure, precision, and recall
Post name/topic F-measure Precision Recall Accuracy
The prime minister and faculty visited the digital economy promotion agency 71.00% 67.00% 75.00% 55.00%
All asian countries go to Agritech startups 80.00% 83.00% 77.00% 67.00%
Startup green house 67.00% 100% 50.00% 50.00%
Only 10k for a smart farm!! 78.00% 100% 64.00% 64.00%
Traditional farmers might work harder 71.00% 85.00% 61.00% 55.00%
Startup! Use leaves to make containers 74.00% 78.00% 70.00% 58.00%
Listen field: algorithm helps to form a new agriculture 75.00% 67.00% 86.00% 60.00%
Digital startup exhibition (digital Thailand big bang: digital station) 77.00% 65.00% 96.00% 63.00%
Startup, homegrown vegetable in container carriers 55.00% 50.00% 60.00% 38.00%
Containers from leaves 83.00% 100% 89.00% 74.00%
Soil scanner 88.00% 85.00% 92.00% 79.00%
Chula startup, Covid-19 vaccine from protein in tobacco leaves 84.00% 80.00% 89.00% 73.00%
Overall 75.00% 80.00% 75.00% 61.33%
4.2. Comments separated by post topic
Table 5 reveals that most of the comments were positive except for some comments in a digital
startup exhibition (digital Thailand big bang: digital station). However, it contained negative comments since
the content was irrelevant to the startups. Figure 7 shows the percentages of sentiment analysis, where
positive sentiment (63.70%) was more prevalent than negative sentiment (36.30%).
Table 5. Sentiment analysis by post name/topic
Post name/topic Comment count Pos sentiment Neg sentiment
The prime minister and faculty visited the digital economy promotion agency 61 (100%) 37 (60.66%) 24 (39.34%)
All Asian countries go to Agritech startups 19 (100%) 16 (84.21%) 3 (15.79%)
Startup green house 43 (100%) 26 (60.47%) 17 (39.53%)
Only 10k for a smart farm!! 41 (100%) 29 (70.73%) 12 (29.27%)
Traditional farmers might work harder 16 (100%) 14 (87.5%) 2 (12.5%)
Startup! use leaves to make containers 27 (100%) 17 (62.96%) 10 (37.04%)
ListenField: Algorithm helps to form a new agriculture 26 (100%) 17 (65.38%) 9 (34.62%)
Digital startup exhibition (digital Thailand big bang: digital station) 84 (100%) 35 (41.67%) 49 (58.33%)
Startup, homegrown vegetable in container carriers 40 (100%) 29 (72.5%) 11 (27.5%)
Containers from leaves 9 (100%) 8 (88.89%) 1 (11.11%)
Soil scanner 17 (100%) 14 (82.35%) 3 (17.65%)
Chula startup, COVID-19 vaccine from protein in tobacco leaves 22 (100%) 16 (72.73%) 6 (27.27%)
Applications of this study: the application of sentiment analysis in practice regarding the
government and business sector was demonstrated. The case study focused on a policy of supporting and

 ISSN: 2088-8708
2836
developing agritech startups or related social trends. Further, the findings could be applied in a marketing
context, such as assessing customer and entrepreneur behaviors and perceptions, future markets and business
trends, needs, and people's problems expressed in Facebook comments. Thus, sentiment analysis helps
summarize the people's views in a particular context as in explaining marketing phenomena, sales strategies,
attitudes, and behaviors of consumers. In the same way, sentiment analysis reflects the effectiveness of the
government's policy on applying it into practice. Novelty/originality of this study: the approach demonstrated
could be applied in studying consumer behavior, marketing analysis, spread of information level, and
attitudes. The study's model can be used in other cases to provide insightful suggestions.
Figure 7. Percentages of sentiments in posted comments on agritech startup topics
5. CONCLUSION
This study used the naive Bayes classifier algorithm in binary sentiment analysis with python to
analyze Thai Facebook comments regarding feelings toward posts of agritech startups in Thailand. We found
that there were positive and negative feelings, but positive comments were more frequent. However, some
words were slang or unstructured and not possible to analyze using the model, so the researchers eliminating
them at the stage of data selection and data cleaning to reduce errors in the results. In further research, we
would adjust the accuracy of classifying comments or data on feelings from key informants. Consequently,
the algorithm will be able to analyze words and phrases more accurately in Thai language. The findings could
be used in marketing analysis or for national strategy and policy, such as exploring attitudes, opinions,
openness, acceptance, responses to new information, or engagement to benefit the government, business
sector, and the people.
Besides, the sentiment analysis showed that most comments were positive (63.70%). This reflects
people and domestic investors' positive interests, acceptance, attitudes toward agritech startups that are new
and trendy. People pay attention to this trend even the startups are based on knowledge of technology and
innovation. It was also found that these positive reflections reveal a level of perception and attentiveness
toward agritech startups in Thailand, even though it is for specific groups. Moreover, the positive views also
show good engagement and creative critiques.
In contrast, negative comments were 36.30%. It might be a view related to temporary businesses. A
large sum of investment funds is another factor driving Agritech startups besides technology, innovation, and
knowledge. Therefore, this factor might affect people's opinions negatively. However, the study will be the
foundation for assessing agritech startup trends in Thailand, such as consumer behavior and marketing
analysis.
Limitations of this study: regarding the search and collection of Facebook posts, the limitations are
as follows: i) there was a small number of agritech startup related posts because this is a niche market and too
specific, so not many people are interested; and ii) since it is too specific and it is in Thai, engagement rate is
low compared to English or Chinese language posts further research: the researchers plan to use sentiment
analysis to determine agritech startup entrepreneurs' opinions in southern Thailand in combination with in-
depth interviews. The second phase is to analyze in-depth opinions regarding the use of technology to inspect
agritech startups and factors influencing the success of using technology.
63.70%
36.30%
Pos sentiment Neg sentiment

2837
ACKNOWLEDGEMENTS
The authors thank Seppo J. Karrila for constructive comments on the manuscript. The work was
supported by Prince of Songkla University, Surat Thani Campus and Graduate School, Prince of Songkla
University.
REFERENCES
[1] A. Baj-Rogowska, “Sentiment analysis of Facebook posts: the uber case,” in 2017 Eighth International Conference on Intelligent
Computing and Information Systems (ICICIS), Dec. 2017, pp. 391–395, doi: 10.1109/INTELCIS.2017.8260068.
[2] S. Akter and M. T. Aziz, “Sentiment analysis on facebook group using lexicon based approach,” in 3rd International Conference
on Electrical Engineering and Information Communication Technology (ICEEICT), Sep. 2016, pp. 1–4, doi:
10.1109/CEEICT.2016.7873080.
[3] C. Troussas, M. Virvou, K. J. Espinosa, K. Llaguno, and J. Caro, “Sentiment analysis of Facebook statuses using Naive Bayes
classifier for language learning,” in Information, Intelligence, Systems and Applications (IISA), Jul. 2013, pp. 1–6, doi:
10.1109/IISA.2013.6623713.
[4] B. Pang and L. Lee, Opinion mining and sentiment analysis, vol. 2, no. 1–2. now Publishers Inc, 2008.
[5] A. Tumasjan, T. O. Sprenger, P. G. Sandner, and I. M. Welpe, “Predicting elections with Twitter: what 140 characters reveal
about political sentiment,” ICWSM 2010 - Proceedings of the 4th International AAAI Conference on Weblogs and Social Media,
pp. 178–185, 2010.
[6] A. Bermingham, M. Conway, L. McInerney, N. O’Hare, and A. F. Smeaton, “Combining social network analysis and sentiment
analysis to explore the potential for online radicalisation,” in 2009 International Conference on Advances in Social Network
Analysis and Mining, Jul. 2009, pp. 231–236, doi: 10.1109/ASONAM.2009.31.
[7] B. Liu, “Sentiment analysis and opinion mining,” Synthesis Lectures on Human Language Technologies, vol. 5, no. 1, pp. 1–167,
May 2012, doi: 10.2200/S00416ED1V01Y201204HLT016.
[8] H. P. Luhn, “A business intelligence system,” IBM Journal of Research and Development, vol. 2, no. 4, pp. 314–319, Oct. 1958,
doi: 10.1147/rd.24.0314.
[9] A.-K. Al-Tamimi, A. Shatnawi, and E. Bani-Issa, “Arabic sentiment analysis of YouTube comments,” in 2017 IEEE Jordan
Conference on Applied Electrical Engineering and Computing Technologies (AEECT), Oct. 2017, pp. 1–6, doi:
10.1109/AEECT.2017.8257766.
[10] M. Hammad and M. Al-awadi, “Sentiment analysis for Arabic reviews in social networks using machine learning,” in Advances
in Intelligent Systems and Computing, vol. 448, Springer International Publishing, 2016, pp. 131–139.
[11] S. Kemp, “Digital 2020: global digital overview,” Datareportal, 2021. https://datareportal.com/reports/digital-2020-global-
digital-overview (accessed May 05, 2021).
[12] P. Vateekul and T. Koomsubha, “A study of sentiment analysis using deep learning techniques on Thai Twitter data,” in 2016
13th International Joint Conference on Computer Science and Software Engineering (JCSSE), Jul. 2016, pp. 1–6, doi:
10.1109/JCSSE.2016.7748849.
[13] T. Chumwatana, “Using sentiment analysis technique for analyzing Thai customer satisfaction from social media,” in
Proceedings of the 5th International Conference on Computing and Informatics, ICOCI 2015, 2015, no. 146, pp. 659–664.
[14] K. Pasupa, P. Netisopakul, and R. Lertsuksakda, “Sentiment analysis of Thai children stories,” Artificial Life and Robotics, vol.
21, no. 3, pp. 357–364, Sep. 2016, doi: 10.1007/s10015-016-0283-8.
[15] N. Kewsuwun, K. Kwiecien, and C. Sae-Chan, “Research knowledge management system on problems, needs and strategies of
Southern Thailand,” Humanities and Social Sciences Reviews, vol. 8, no. 1, pp. 323–333, Jan. 2020, doi:
10.18510/hssr.2020.8143.
[16] S. Gusriani, K. D. K. Wardhani, and M. I. Zul, “Sentiment analysis of online stores in social media using the Naive Bayes
classification method (Case Study: BerryBenka Facebook Page). [in Bahasa],” 4th Applied Business and Engineering Conference,
vol. 1, no. 1. pp. 1–7, 2016.
[17] A. B. Pawar, M. A. Jawale, and D. N. Kyatanavar, “Fundamentals of sentiment analysis: concepts and methodology,” in Studies
in Computational Intelligence, vol. 639, Springer International Publishing, 2016, pp. 25–48.
[18] T. Nasukawa and J. Yi, “Sentiment analysis: capturing favorability using natural language processing,” in Proceedings of the 2nd
International Conference on Knowledge Capture, K-CAP 2003, 2003, pp. 70–77, doi: 10.1145/945645.945658.
[19] S. Y. Ganeshbhai and B. K. Shah, “Feature based opinion mining: A survey,” in Souvenir of the 2015 IEEE International
Advance Computing Conference, IACC 2015, 2015, pp. 919–923, doi: 10.1109/IADCC.2015.7154839.
[20] N. Jindal and B. Liu, “Mining comparative sentences and relations,” in AAAI’06 Proceedings of the 21st National Conference on
Artificial Intelligence, 2006, vol. 2, pp. 1331–1336.
[21] A. Moreo, M. Romero, J. L. Castro, and J. M. Zurita, “Lexicon-based comments-oriented news sentiment analyzer system,”
Expert Systems with Applications, vol. 39, no. 10, pp. 9166–9180, Aug. 2012, doi: 10.1016/j.eswa.2012.02.057.
[22] W. Medhat, A. Hassan, and H. Korashy, “Sentiment analysis algorithms and applications: a survey,” Ain Shams Engineering
Journal, vol. 5, no. 4, pp. 1093–1113, Dec. 2014, doi: 10.1016/j.asej.2014.04.011.
[23] N. A. Vidya, M. I. Fanany, and I. Budi, “Twitter sentiment to analyze net brand reputation of mobile phone providers,” Procedia
Computer Science, vol. 72, pp. 519–526, 2015, doi: 10.1016/j.procs.2015.12.159.
[24] E. Cambria, A. Hussain, and C. Havasi, “Towards the Crowd Validation of the British National Health Service,” Proceedings of
the WebSci10: Extending the Frontiers of Society On-Line, April 26-27th, 2010, Raleigh, NC: US. Webscience Trust, 2010.
[25] A. Bakliwal, J. Foster, J. van der Puil, R. O’Brien, L. Tounsi, and M. Hughes, “Sentiment analysis of political tweets: towards an
accurate classifier,” in Proceedings of the Workshop on Language Analysis in Social Media, 2013, pp. 49–58.
[26] L. Banić, A. Mihanović, and M. Brakus, “Using big data and sentiment analysis in product evaluation,” in Information and
Communication Technology Electronics ang Microelectronics (MIPRO), 2013 36th International Convention, 2013, pp. 1149–
1154.
[27] N. A. M. Zamani, S. Z. Z. Abidin, N. Omar, and M. Z. Z. Abiden, “Sentiment analysis: determining people’s emotions in
Facebook,” in Proceedings of the 13th International Conference on Applied Computer and Applied Computational Science, 2014,
pp. 111–116.
[28] B. Haryanto, Y. Ruldeviyani, F. Rohman, J. D. T.N., R. Magdalena, and Y. F. Muhamad, “Facebook analysis of community
sentiment on 2019 Indonesian presidential candidates from Facebook opinion data,” Procedia Computer Science, vol. 161, pp.
715–722, 2019, doi: 10.1016/j.procs.2019.11.175.

 ISSN: 2088-8708
2838
[29] P. Songram and C. Jareanpon, “Opinion mining of Thai politics on Facebook status updates,” in Proceedings of The 3rd
International Conference on Intelligent Systems and Image Processing 2015, 2015, pp. 240–245, doi: 10.12792/icisip2015.046.
[30] J. Akaichi, Z. Dhouioui, and M. J. Lopez-Huertas Perez, “Text mining Facebook status updates for sentiment classification,” in
2013 17th International Conference on System Theory, Control and Computing (ICSTCC), Oct. 2013, pp. 640–645, doi:
10.1109/ICSTCC.2013.6689032.
[31] B. Pang, L. Lee, and S. Vaithyanathan, “Thumbs up? sentiment classification using machine learning techniques,” in Proceedings
of the 2002 Conference on Empirical Methods in Natural Language Processing (EMNLP 2002), 2002, vol. 10, no. 4, pp. 79–86,
doi: 10.3115/1118693.1118704.
[32] C.-H. Lee, “An information-theoretic filter approach for value weighted classification learning in Naive Bayes,” Data &
Knowledge Engineering, vol. 113, pp. 116–128, Jan. 2018, doi: 10.1016/j.datak.2017.11.002.
[33] R. Huggins and A. Johnston, “Knowledge networks in an uncompetitive region: SME innovation and growth,” Growth and
Change, vol. 40, no. 2, pp. 227–259, Jun. 2009, doi: 10.1111/j.1468-2257.2009.00474.x.
[34] “Online meeting report of the competitive of business and human,” Bangkok: Kenan Institute Asia. Bangkok, 2015.
[35] “Situation of small and medium enterprises 2018 report,” Office of Policy and Planning. Bangkok, 2018.
[36] A. Bromberg and A. Lima, “Sentiment analysis in Python,” GitHub, 2015.
https://github.com/abromberg/sentiment_analysis_python (accessed May 05, 2021).
[37] T. H. Soliman, M. A. Elmasry, A. Hedar, and M. M. Doss, “Sentiment analysis of arabic slang comments on Facebook,”
International Journal of Computers and Technology, vol. 12, no. 5, pp. 3470–3478, Jan. 2014, doi: 10.24297/ijct.v12i5.2917.
BIOGRAPHIES OF AUTHORS
Nawapon Kewsuwun received the B.B.A. degree in computer information
system from Rajamangala University of Technology Tawan-Ok, Thailand, in 2012 and the
M.I.ED. Degrees in Educational Technology from King Mongkut’s Institute of Technology
Ladkrabang, Thailand, in 2015. Currently, he is Lecturer at the Department of Information
Management, Faculty of Humanities and Social Sciences, Prince of Songkla University,
Pattani Campus. His research interests include Information Science, Information and
Knowledge Management, Technology administration, and Educational technology adoption
and Utilization. He can be contacted at email: nawapon.k@psu.ac.th.
Siriwan Kajornkasirat received the Ph.D. degree in Computational Science from
Walailak University, Thailand, in 2011. She has done on Ph.D. research experience in Deakin
University, Australia funded by the Royal Golden Jubilee Ph.D. Program (RGJ-Ph.D.
Program). In 2014, she was invited for STEM Education workshop under the International
Visitor Leadership Program (IVLP). This is a program of the U.S. Department of State with
funding provided by the U.S. Government. Currently, she is Assistant Professor at the Faculty
of Science and Industrial Technology, Prince of Songkla University, Surat Thani, Thailand.
Her research interests include Data Science, Computing Science, Advanced Analytic Online,
STEM Education, Smart Farming, Internet of Things (IoT), Smart Health, Digital Marketing,
E-Marketing for Tourism. She can be contacted at email: siriwan.wo@psu.ac.th.

A sentiment analysis model of agritech startup on Facebook comments using naive Bayes classifier

More Related Content

Similar to A sentiment analysis model of agritech startup on Facebook comments using naive Bayes classifier (20)

More from IJECEIAES (20)

Recently uploaded (20)

A sentiment analysis model of agritech startup on Facebook comments using naive Bayes classifier