KMITL

Permanent URI for this communityhttps://dspace.kmitl.ac.th/handle/kmitl/1

Browse

Search Results

Now showing 1 - 5 of 5

Some of the metrics are blocked by your
consent settings
Aspect-Level Sentiment Analysis Using WangchanBERTa for Fine-Grained Service Insight Extraction in Hotel Reviews
(2026-06-01)
Suwan, Thanachok
;
Nokkaew, Manussawee
;
Surawanitkun, Chayada
;
Sorn-In, Kanda
;
Mueanrit, Nongram
Online booking site reviews substantially influence Thai SME hotel reputations and consumer decisions. Hotels should readily extract useful information from unstructured Thai-language ratings. WangchanBERTa, a Thai deep learning model, automates hotel sentiment analysis and strategic insight development in this study. System is two-stage. Phase 1 divides 10,040 Thai hotel reviews from Agoda, Booking.com, Traveloka, and Trip.com into good and negative attitudes and determines price, service quality, and cleanliness. Phase 2 extracts aspect-level information across 11 service characteristics to discover complex trends like consumers being satisfied with service but unhappy with cost. The sentiment categorization model performed well with 91.63% accuracy and 89.69% macro F1-score in experiments. The aspect-based sentiment analysis system achieved 91.63% accuracy, 91.55% macro precision, 91.63% recall, 91.52% F1-score, and real-world insight extraction. This methodology helps hoteliers listen to customers, integrate data into business ideas, and compete in Thailand’s tourism market.
3
Some of the metrics are blocked by your
consent settings
A Comparative Study of NLP-Based Models: Popularity Trends in Pop Mart Comments
(2026-05-04)
Charoenporn, Pattama
;
Poonsawat, Kititinun
;
Banjongkan, Anupong
In today's digital era, where consumer sentiment significantly influences brand perception, understanding customer emotions has become a strategic priority for companies like POP MART. Among its iconic collectibles, Labubu stands out as a symbol of the brand's global appeal, gaining tremendous popularity through the viral influence of Lisa from BLACKPINK, who showcased her Labubu collection on social media. This moment sparked a surge in interest, especially in Thailand, where Labubu achieved over 365,000 mentions in a single month, driven by its appeal to young, trend-conscious consumers. This study, Sentiment Analysis of LABUBU Popularity Trends: A Study of POPMART Comments Using NLP Techniques, investigates customer perceptions and emotions toward Labubu by employing advanced transformer-based NLP models - WangchanBERTa, RoBERTa, and XLM-R. By analyzing Thai-specific and multilingual datasets collected from social media platforms, this research evaluates the models' performance in sentiment classification. The analysis highlights WangchanBERTa's superior performance on Thai-specific datasets, achieving the highest F1-score of 87.51%. XLM-R demonstrated robust multilingual capabilities with an F1-score of 84.65%, while RoBERTa excelled in English sentiment analysis with an F1-score of 82.83%. These findings emphasize the importance of leveraging both language-specific and multilingual models to address cultural and linguistic nuances effectively. Sentiment patterns reveal positive comments highlighting emotional connections to Labubu, neutral sentiments about logistical inquiries, and negative feedback on technical and availability issues. These insights guide strategies to enhance marketing, boost customer engagement, and address operational challenges. This study advances NLP applications in sentiment analysis, offering POP MART data-driven insights to refine global and regional strategies while underscoring the value of multilingual transformer models in analyzing diverse consumer feedback.
1
Some of the metrics are blocked by your
consent settings
ENHANCING PLATFORM CREDIBILITY IN THAI E-COMMERCE: RECALIBRATING INFLATED STAR RATINGS VIA SENTIMENT ANALYSIS
(2026-01-01)
Leenawong, Chartchai
;
Thangthawornkit, Warunada
;
Ritthipakdee, Amarita
This study addresses the erosion of platform credibility in international e-commerce caused by the misalignment between numerical star ratings and actual textual sentiment. While star ratings are a dominant mechanism guiding purchasing decisions in rapidly expanding markets like Southeast Asia, they are frequently inflated due to platform-induced biases and region-specific cultural factors, such as the politeness norms and indirect communication strategies prevalent in Thai culture. To overcome this structural limitation, this research aims to empirically quantify this discrepancy and introduce a transparent, sentiment-informed metric—the "Heart Score"—to recalibrate ratings. A rule-based sentiment analysis framework specifically adapted for Thai-language reviews was applied to a dataset of 14,877 women’s clothing entries from Lazada. By leveraging expert-validated lexicons and linguistic rules for negation and amplification, store-level sentiment ratios were extracted and transformed into recalibrated scores via a normalized linear function with penalty adjustments. Findings confirm substantial sentiment misalignment: although the analyzed stores clustered narrowly with star ratings between 4.92 and 5.00, the recalibrated Heart Scores spanned a significantly wider, more meaningful range of 3.81 to 4.71. The most notable gap of 1.11 points exposed hidden dissatisfaction masked by near-perfect ratings. The proposed framework offers strong managerial insights for platform governance by enhancing review credibility, enabling more accurate product benchmarking, and providing a scalable, interpretable model for restoring trust in low-resource language environments.
Some of the metrics are blocked by your
consent settings
Thai Sentiment Analysis via Bidirectional LSTM-CNN Model with Embedding Vectors and Sentic Features
(2018-07-02)
Ayutthaya, Thititorn Seneewong Na
;
Pasupa, Kitsuchart
Sentiment analysis is one of the most frequently performed tasks in Natural Language Processing that plays an important role in marketing research. It allows us to understand customer sentiment. The outcomes from this kind of analysis can be used to improve products and services. Recently, a Word2Vec model, a technique for word embedding (converting text into a number) has been developed and used successfully to a degree to get the sentiment of customers from the text responses that they provided. This work attempted to incorporate two more features-part-of-speech and sentic features-to make the analysis more accurate. The part-of-speech feature identifies the type of words that better convey various sentiments, while the sentic feature identifies the emotion underlying certain words. Combining Bidirectional Long Short-term Memory and Convolutional Neural Networks models with several combinations of the features mentioned, we performed a sentiment analysis of Thai children stories and found that the combination of all three features gave the best result at 78.89 % F1-score.
1
Some of the metrics are blocked by your
consent settings
Social commerce credibility analysis
(2017-02-23)
Kraichit, Keeta
;
Amarittakul, Jantharat
;
Netisopakul, Ponrudee
;
Thanasopon, Bundit
Online shopping is increasingly popular for customers. However, there are also high concerns on online frauds. In order to alleviate this concern, a proposed Social Commerce pages' Credibility Analysis software (SCCA) is developed. The web application can provide up-To-date information related to the user's chosen social commerce pages on Facebook. It provides figures and statistics that indicate the page's credibility. In addition, it also provides sentiment scores, both positive and negative, which are analyzed based on customers' comments on the page. The results from using this SCCA software on two social commerce categories, i.e. fashion and IT-each with ten popular pages, are used to rank these pages. The ranking is done based on our own devising measurement. This gives us some more insights into the current de-facto standard of common believed credibility. For example, it suggests that the amount of 'Like', which is often used to assess a page's credibility, may not be adequate for a complex issue such as social commerce credibility.

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results