Pusadee Seresangtakul

Dr. Pusadee Seresangtakul

Department of Computer Science, Khon Kaen University
orcid

Research Focus & Metrics

Academic Career Length

25Years (2002-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
13 May 2023 in Data
Verified
Journal Article
Open access
PDF file
NKNattawat Khanmphakdee
Nattawat Khanmphakdee
PSPusadee Seresangtakul
Pusadee Seresangtakul
The number of reviews from customers on travel websites and platforms is quickly increasing. They provide people with the ability to write reviews about their experience with respect to service quality, location, room, and cleanliness, thereby helping others before booking hotels. Many people fail to consider hotel bookings because the numerous reviews take a long time to read, and many are in a non-native language. Thus, hotel businesses need an efficient process to analyze and categorize the polarity of reviews as positive, negative, or neutral. In particular, low-resource languages such as Thai have greater limitations in terms of resources to classify sentiment polarity. In this paper, a sentiment analysis method is proposed for Thai sentiment classification in the hotel domain. Firstly, the Word2Vec technique (the continuous bag-of-words (CBOW) and skip-gram approaches) was applied to create word embeddings of different vector dimensions. Secondly, each word embedding model was combined with deep learning (DL) models to observe the impact of each word vector dimension result. We compared the performance of nine DL models (CNN, LSTM, Bi-LSTM, GRU, Bi-GRU, CNN-LSTM, CNN-BiLSTM, CNN-GRU, and CNN-BiGRU) with different numbers of layers to evaluate their performance in polarity classification. The dataset was classified using the FastText and BERT pre-trained models to carry out the sentiment polarity classification. Finally, our experimental results show that the WangchanBERTa model slightly improved the accuracy, producing a value of 0.9225, and the skip-gram and CNN model combination outperformed other DL models, reaching an accuracy of 0.9170. From the experiments, we found that the word vector dimensions, hyperparameter values, and the number of layers of the DL models affected the performance of sentiment classification. Our research provides guidance for setting suitable hyperparameter values to improve the accuracy of sentiment classification for the Thai language in the hotel domain.

1 February 2017
Verified
Conference Paper
Closed access
TPThongpan Pariwat
Thongpan Pariwat
PSPusadee Seresangtakul
Pusadee Seresangtakul
This paper presents a finger-spelling recognition system focusing on Thai finger-spelling sign language, derived from the computer vision, using SVM. In this study, global and local features were extracted from input finger images. In order to develop the recognition system, 15 Thai alphabet characters were collected from five hand signers, totally 375 character pictures, in order to train the system using the SVM technique; with linear, polynomial, RBF, and sigmoid kernels. Each kernel method employed three feature vectors extracted from global features, local features, and the combination of both features; and were measured for performance in 4 SVM kernels, with five-fold cross-validation. The experimental results demonstrated that the combination of global and local features applied in RBF, linear, polynomial, and sigmoid resulted in the average accuracies of 91.20%, 86.40%, 80.00%, and 54.67%, respectively. The RBF method with the combination of global and local features provided the highest accuracy among all combinations.

4 February 2021 in Symmetry
Verified
Journal Article
Open access
PDF file
TPThongpan Pariwat
Thongpan Pariwat
PSPusadee Seresangtakul
Pusadee Seresangtakul
Sign language is a type of language for the hearing impaired that people in the general public commonly do not understand. A sign language recognition system, therefore, represents an intermediary between the two sides. As a communication tool, a multi-stroke Thai finger-spelling sign language (TFSL) recognition system featuring deep learning was developed in this study. This research uses a vision-based technique on a complex background with semantic segmentation performed with dilated convolution for hand segmentation, hand strokes separated using optical flow, and learning feature and classification done with convolution neural network (CNN). We then compared the five CNN structures that define the formats. The first format was used to set the number of filters to 64 and the size of the filter to 3 × 3 with 7 layers; the second format used 128 filters, each filter 3 × 3 in size with 7 layers; the third format used the number of filters in ascending order with 7 layers, all of which had an equal 3 × 3 filter size; the fourth format determined the number of filters in ascending order and the size of the filter based on a small size with 7 layers; the final format was a structure based on AlexNet. As a result, the average accuracy was 88.83%, 87.97%, 89.91%, 90.43%, and 92.03%, respectively. We implemented the CNN structure based on AlexNet to create models for multi-stroke TFSL recognition systems. The experiment was performed using an isolated video of 42 Thai alphabets, which are divided into three categories consisting of one stroke, two strokes, and three strokes. The results presented an 88.00% average accuracy for one stroke, 85.42% for two strokes, and 75.00% for three strokes.

22 December 2016 in BMC Genomics
Verified
Journal Article
Open access
PDF file
PPPiyarat Ponyared
Piyarat Ponyared
JPJiradej Ponsawat
Jiradej Ponsawat
STSissades Tongsima
Sissades Tongsima
PSPusadee Seresangtakul
Pusadee Seresangtakul
CAChutipong Akkasaeng
Chutipong Akkasaeng
NTNathpapat Tantisuwichwong
Nathpapat Tantisuwichwong
Simple sequence repeats (SSRs) have become widely used as molecular markers in plant genetic studies due to their abundance, high allelic variation at each locus and simplicity to analyze using conventional PCR amplification. To study plants with unknown genome sequence, SSR markers from Expressed Sequence Tags (ESTs), which can be obtained from the plant mRNA (converted to cDNA), must be utilized. With the advent of high-throughput sequencing technology, huge EST sequence data have been generated and are now accessible from many public databases. However, SSR marker identification from a large in-house or public EST collection requires a computational pipeline that makes use of several standard bioinformatic tools to design high quality EST-SSR primers. Some of these computational tools are not users friendly and must be tightly integrated with reference genomic databases. A web-based bioinformatic pipeline, called EST Analysis Pipeline Plus (ESAP Plus), was constructed for assisting researchers to develop SSR markers from a large EST collection. ESAP Plus incorporates several bioinformatic scripts and some useful standard software tools necessary for the four main procedures of EST-SSR marker development, namely 1) pre-processing, 2) clustering and assembly, 3) SSR mining and 4) SSR primer design. The proposed pipeline also provides two alternative steps for reducing EST redundancy and identifying SSR loci. Using public sugarcane ESTs, ESAP Plus automatically executed the aforementioned computational pipeline via a simple web user interface, which was implemented using standard PHP, HTML, CSS and Java scripts. With ESAP Plus, users can upload raw EST data and choose various filtering options and parameters to analyze each of the four main procedures through this web interface. All input EST data and their predicted SSR results will be stored in the ESAP Plus MySQL database. Users will be notified via e-mail when the automatic process is completed and they can download all the results through the web interface. ESAP Plus is a comprehensive and convenient web-based bioinformatic tool for SSR marker development. ESAP Plus offers all necessary EST-SSR development processes with various adjustable options that users can easily use to identify SSR markers from a large EST collection. With familiar web interface, users can upload the raw EST using the data submission page and visualize/download the corresponding EST-SSR information from within ESAP Plus. ESAP Plus can handle considerably large EST datasets. This EST-SSR discovery tool can be accessed directly from: http://gbp.kku.ac.th/esap_plus/.

10 September 2021 in Acta Informatica Pragensia
Verified
Journal Article
Open access
PDF file
NKNattawat Khamphakdee
Nattawat Khamphakdee
PSPusadee Seresangtakul
Pusadee Seresangtakul
Peer-reviewed journal on social and business aspects of informatics that covers the theory and application of enterprise information systems, as well as interactions between ICT and people. Recenzovan? ?asopis o soci?ln?ch a byznys aspektech informatiky zahrnuj?c? teorie a aplikace podnikov?ch informa?n?ch syst?m? a interakce mezi ICT a lidmi.

1 August 2018
Verified
Conference Paper
Closed access
PSPongsakorn Saipech
Pongsakorn Saipech
PSPusadee Seresangtakul
Pusadee Seresangtakul
This paper presents an automatic subjective examination system for Thai language based on semantic similarity, applying cosine similarity techniques together with the applicable synonym. The examination answers are input into the system in short Thai sentences, without word boundaries. The input answers which are reference answers, and student answers, were segmented into a sequence of Thai words; using the longest matching algorithm and dictionary (Lexitron ® ). These answers were transformed into vectors using TF-IDF, which is a frequency computation technique. The synonyms were also considered. Cosine similarity was used to measure the similarity of the answers. In order to evaluate the performance of the proposed system, we prepared a dataset of five questions from the final exam of the `Database Management System and Database Design' course, and asked 70 students to provide their answers. The datasets were prepared for both systems, with and without synonyms, in order to score the student answers. We crosschecked the scores of both systems with an expert, their teacher. In comparing the results of both systems, we found that the scores produced from the system using cosine similarity with synonyms were similar to those obtained from the expert.