KMITL

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

Browse

Search Results

Now showing 1 - 7 of 7

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
Thai Generalized Alignment Task (TGAT): A Corpus and Comparative Study for Hallucination Detection in Thai
(2025-01-01)
Wangwon, Sivakorn
;
Pitijaroonpong, Rapepong
;
Chuangkrud, Piyawat
;
Damrongrat, Chaianun
;
Kongyoung, Sarawoot
Large Language Models (LLMs) face critical reliability challenges due to hallucination-the generation of factually inaccurate content. While hallucination detection has advanced for major languages, Thai remains underserved, lacking specialized alignment models and datasets. We introduce the Thai Generalized Alignment Task (TGAT), a seven-subtask corpus tailored for Thai hallucination detection, spanning Fact Verification, Natural Language Inference, Information Retrieval, Question Answering, Summarization, Semantic Textual Similarity, and Paraphrase Identification. We conduct a comparative study across transformer families-encoder-only, encoder-decoder, and decoder-only-evaluated on the test set of each sub-task to analyze architectural compatibility and generalization for Thai hallucination detection. We also perform an ablation study to quantify how the presence of each subtask dataset affects the overall average performance, clarifying the role of dataset composition. Our experiments show that decoder-only models consistently outperform encoder-decoder and encoder-only alternatives under a standardized zero-shot prompting setup, establishing strong baselines and offering guidance for model selection and dataset design in low-resource settings.
Some of the metrics are blocked by your
consent settings
Visualizing Political Communication Trends across Generations on X (Twitter): Insights Through Topic Modeling and Word Clouds
(2024-01-01)
Udomwisanpat, Prinwat
;
Jitkajornwanich, Kulsawasd
;
Kraishan, Obada
;
Srestasatheirn, Panu
;
Lawawirojwong, Siam
This study examines the interests and significance of words on Twitter (or X) across different generational groups: Baby Boomers, Generation X, Generation Y, and Generation Z. Using Topic Modeling with Latent Dirichlet Allocation (LDA), the research explores relationships and word importance within each group. As the results of topic modeling are not always easy to interpret, we used word cloud visualization to help make sense of the results for each generation. The findings reveal distinct patterns: Baby Boomers frequently mention print media, news websites, and prominent Thai political figures; Generation X emphasizes individuals and local political issues in Bangkok; Generation Y discusses political and social events; and Generation Z uniquely questions political and social activities. This research methodology is applicable across languages and tasks, offering insights into generational behaviors and interests.
1
Some of the metrics are blocked by your
consent settings
Improving OpenAI's Whisper Model for Transcribing Homophones in Legal News
(2024-01-01)
Siriket, Lattapon
;
Jitkajornwanich, Kulsawasd
;
Jaiyen, Saichon
;
Intakosum, Sarun
The 'Whisper' model provides a tool for those who require transcription of human voice. It equips with opensource features and diverse functionalities. The model is capable of effectively deciphering messages in multiple languages, including support for the Thai language. This paper focuses on improving the transcription process of Thai homophones using the Whisper model in reducing the word error rate (WER). We focus on words in the legal news category and identify factors that lead to Whisper's incorrect sound predictions. We examined homophones using snippets of legal news video clips and compiled them into a homophone dictionary. We compare words extracted from the Whisper model by determining the word error rate and spelling of words. Based on the initial results obtained from the original Whisper model and the created homophone dictionary, 48 % of the words were incorrectly transcribed out of a total of 94 words. Then, we propose a methodology by which the performance of the Whisper is improved. That way, the automatic speech recognition of Thai language using the Whisper model can fully be utilized and used in other applications.
Some of the metrics are blocked by your
consent settings
Hate Speech Detection in Thai Social Media with Ordinal-Imbalanced Text Classification
(2022-01-01)
Pasupa, Kitsuchart
;
Karnbanjob, Werasut
;
Aksornsiri, Massakorn
Cyberbullying has become a serious problem in Thai social media. For example, some Thai people posted hate speeches on Myanmar workers in Thailand during the COVID-19 pandemic, which might elevate hate crime. It is imperative and urgent to detect cyberbullying on Thai social media. The task is a text classification problem. Moreover, hate speeches contain the order of severity levels, but many pieces of work did not consider this point in the model. Therefore, we developed a Thai hate-speech classification method with various loss functions to detect such hate speeches accurately. We evaluated them on a corpus of ordinal-imbalanced Thai text. The evaluated outcomes indicated that the best-in terms of $F$1 -score-model was the model with a loss function of a hybrid between an Ordinal regression loss function and Pearson correlation coefficients (common in similarity function). It yielded an average F1-score of 78.38 %-0.88 % significantly higher than the score achieved by a conventional loss function-and an average mean squared error of 0.2478-5.49 % relative improvement. Thus, the proposed hybrid loss function improved the efficiency of the model.
1
Some of the metrics are blocked by your
consent settings
Improved Identification of Imbalanced Multiple Annotation Intent Labels with a Hybrid BLSTM and CNN Model and Hybrid Loss Function
(2021-01-01)
Vatathanavaro, Supawit
;
Pasupa, Kitsuchart
;
Sirirattanajakarin, Sorratat
;
Suntisrivaraporn, Boontawee
Payment or fund transfer transactions can be annotated by users when they are made through a mobile banking app, for example, SCB Easy app—a mobile banking app by Siam Commercial Bank—allows users to annotate transactions with 40 character texts. The AI<sup>2</sup> framework was used to identify user intentions with the transactions, so that the bank can offer the right product to the right customer at the right time. The framework employed Long Short-Term Memory (LSTM). Commonly, one annotated sample can be interpreted as representing multiple intents, thus we had a multiple label classification problem. However, the original model did not consider the class imbalance, that caused the model to bias toward the majority class. We introduced a new hybrid Bidirectional LSTM and Convolutional Neural Network model in conjunction with a new hybrid loss function to tackle the imbalance. Our model with hybrid loss function performed better than the AI<sup>2</sup> framework with a 4.5% improvement in F<inf>1</inf> -score. Moreover, our hybrid loss function enabled the model to classify minority classes better, when the imbalance ratio became higher, compared with a conventional cross-entropy loss function. In other words, our hybrid loss function made the model to be more efficient in real-world multiple label imbalance problem.
Some of the metrics are blocked by your
consent settings
Survey of Query correction for Thai business-oriented information retrieval
(2020-11-18)
Kittiworapanya, Phongsathorn
;
Saelek, Nuttapong
;
Lertpiya, Anuruth
;
Chalothorn, Tawunrat
The importance of effective Thai information retrieval (IR) increases as more businesses in Thailand undergo digital transformation. However, previous research on Thai IR systems has mainly focused on web search engines. This study will focus on using query correction to reduce user errors to improve Thai IR. Experiments are conducted on our business-oriented Thai IR task (bTIR). Our investigation presented three notable findings. First, cognitive errors are less of an issue in a business setting. Thus, homophones correction methods provide very little to no benefit for bTIR. Second, approximation based spelling correction methods can significantly reduce search performance. Thus, partial matching on a full dictionary, such as symmetric delete indexing (SymSpell), should be preferred over non-optimal search methods. Third, we introduce a re-ranking algorithm for query corrector, which features multiple sub-correctors (e.g., ThaiQCor 2.0), which results in better performance across multiple configurations.

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results