Panida Songram

Thai sentiment analysis on social media remains challenging due to the lack of explicit word boundaries, informal expressions, code-mixing, short text length, and severe class imbalance, particularly in the WISESIGHT benchmark dataset. Although transformer-based models such as WangchanBERTa provide strong contextual representations for Thai language processing, conventional hybrid architectures using fixed convolutional kernels may fail to capture diverse local semantic patterns across different sentiment classes. To address this limitation, this study proposes WangchanBERTa with Instance-Class-Aware Dynamic Kernel Selection (WangchanBERTa-IC-DKS), a novel hybrid framework that integrates contextualized embeddings with multi-kernel convolutional neural networks and dynamic kernel selection at both the instance and class levels. The proposed model adaptively determines kernel importance according to sentence semantics and class-specific sentiment characteristics, enabling more effective local feature extraction and improved minority-class recognition. Experiments were conducted on the WISESIGHT sentiment benchmark using four kernel candidates: [2, 3, 4], [3, 4, 5], [2, 3, 4, 5], and [1, 2, 3, 4, 5]. The kernel candidate [2, 3, 4, 5] achieves the best performance, with a mean macro-F1 of 63.93% averaged across three independent runs using different random seeds. Ablation studies further show that combining both class-aware and instance-aware dynamic kernel selection improves balanced classification performance compared with fixed multi-kernel CNN baselines. The proposed model also outperforms the previous state-of-the-art Parallel Hybrid model by 1.23 percentage points on the WISESIGHT dataset and by 1.36 percentage points on the 40 Thai children’s tales dataset. These findings demonstrate that dynamic kernel selection is more effective than fixed convolutional kernels across the two evaluated Thai sentiment benchmarks, particularly for short, noisy, and imbalanced Thai texts.

Sentiment analysis of Thai social media texts remains challenging due to the lack of explicit word boundaries, informal language, and high linguistic variability, further exacerbated by class imbalance, short texts, and ambiguous categories. Recent transformer-based models, particularly WangchanBERTa, have shown strong performance in Thai NLP; however, their ability to extract fine-grained, sentiment-specific local features remains limited when applied to noisy, short social media texts. This study proposes a lightweight hybrid framework that integrates WangchanBERTa with multiple neural architectures, including convolutional neural networks (CNN), bidirectional long short-term memory (BiLSTM), and bidirectional gated recurrent units (BiGRU). A comparative evaluation is conducted to assess their effectiveness for four-class Thai sentiment classification. In addition, the impact of CNN kernel size configurations on sentiment feature extraction is systematically investigated, an aspect that has received limited attention in prior Thai NLP research. Experiments conducted on the WISESIGHT benchmark demonstrate that the WangchanBERTa–CNN model with medium kernel sizes [2, 3, 4] achieves the best performance. It achieves a macro-average F1-score of 65.80%, outperforming the current state-of-the-art on the WISESIGHT benchmark by 4.94%. In addition, it achieves a macro-average F1-score of 94.11% on the class-balanced dataset. These results are achieved while maintaining a relatively low parameter count. These findings confirm that combining contextualized global embeddings with local n-gram feature extraction provides an effective and efficient solution for sentiment classification of short, noisy, and linguistically diverse Thai social media texts.

Future programmable networks such as 5G/6G and large-scale IoT deployments demand dynamic and intelligent bandwidth control mechanisms to ensure stable Quality of Service (QoS) under highly variable traffic conditions. Conventional queue-based schedulers and emerging machine learning techniques still struggle with slow reaction to congestion, unstable fairness, and high computational costs. To address these challenges, this paper proposes a Dynamic Predictive Feedback (DPF) mechanism that integrates clustered-LSTM based short-term traffic prediction with meta-control driven adaptive bandwidth adjustment in a Software-Defined Networking (SDN) architecture. The prediction module proactively estimates future queue depth and arrival rates using in-band network telemetry (INT), while the feedback controller continuously adjusts scheduling weights based on congestion risk and fairness metrics. Extensive emulation experiments conducted under Static, Bursty IoT, Mixed, and Stress workloads show that DPF consistently outperforms state-of-the-art solutions, including A-WFQ and DRL-based schedulers, achieving up to 32% higher throughput, up to 40% lower latency, and 10–12% lower CPU and memory usage. Moreover, DPF demonstrates strong fairness (Jain’s Index ≥ 0.96), high adaptability, and minimal performance variance across scenarios. These results confirm that DPF is a scalable and resource-efficient solution capable of supporting the demands of future programmable, 5G/6G-ready network infrastructures.

This research aims to enhance the detection of sarcastic messages in the Thai language across social networks. It involves extracting and analyzing context-based features from messages to identify and differentiate sarcastic content. The study employs deep learning and machine learning techniques to classify these messages. The experimental findings demonstrate that a combination of context-based and content-based features yields the highest accuracy in identification. Specifically, the utilization of a bidirectional Long-Short Term Memory (Bi-LSTM) with 256 nodes, ReLU as the activation function, a dropout rate of 0.2, Sigmoid as the output activation function, binary cross-entropy as the loss function, and the Adam optimizer resulted in the highest accuracy achieved by the Bi-LSTM model, reaching 96.79%.

Online deception has become a major problem affecting people, society, the economy, and national security. It is mostly done by spreading deceptive messages because message are quickly spread on social networks and are easily accessed by anyone. Detecting deceptive messages is challenging as the messages are unstructured, informal, and complex; this extends into Thai language messages. In this paper, various deep learning models are proposed to detect deceptive messages under two feature extraction trials. A balanced two-class dataset of deceptive and truthful Thai messages (n=2378) is collected from Facebook pages. Instance features are encoded using word embeddings (Thai2Fit) and one-hot encoding techniques. Five classification models, convolutional neural network (CNN), bidirectional long short-term memory (BiLSTM), bidirectional gated recurrent units (BiGRU), CNN-BiLSTM, and CNN-BiGRU, are proposed and evaluated upon the dataset with each feature extraction technique. The experimental results show that all the proposed models had excellent accuracy (95.59% to 98.74%) and BiLSTM with one-hot encoding gave the best performance, achieving 98.74% accuracy.

Numerous research have been done to predict cryptocurrency prices since cryptocurrency prices affect global economic and monetary systems. However, investigations using linear connection approaches and technical analysis indicators frequently fall short of providing an explanation for changes in the pattern of BitCoin pricing. This paper is proposed to study time-varying parameters with long short-term memory (LSTM). The study is investigated on a dataset retrieved from Binance from March 2022 to April 2022. The proposed LSTM used a variety of hyperparameter settings, particularly time parameters, to predict the cryptocurrency price (BTC/USDT) on the dataset. Additionally, it is evaluated in terms of mean absolute percentage error (MAPE) in comparison to smooth moving average (SMA), weighted moving average (WMA), and exponential moving averages (EMA). From the investigation, using the previous 3 days for prediction gives the lowest of the MAPE values and the proposed LSTM outperformed the other models. When considering the last three days' value of pricing, the indicated LSTM offers the best accurate prediction, with a MAPE percentage of 0.0927%.

Associative classification (AC) is a mining technique that integrates classification and association rule mining to perform classification on unseen data instances. AC is one of the effective classification techniques that applies the generated rules to perform classification. In particular, the number of frequent ruleitems generated by AC is inherently designated by the degree of certain minimum supports. A low minimum support can potentially generate a large set of ruleitems. This can be one of the major drawbacks of AC when some of the ruleitems are not used in the classification stage, and thus (to reduce the rule-mapping time), they are required to be removed from the set. This pruning process can be a computational burden and massively consumes memory resources. In this paper, a new AC algorithm is proposed to directly discover a compact number of efficient rules for classification without the pruning process. A vertical data representation technique is implemented to avoid redundant rule generation and to reduce time used in the mining process. The experimental results show that the proposed algorithm archives in terms of accuracy a number of generated ruleitems, classifier building time, and memory consumption, especially when compared to the well-known algorithms, Classification-based Association (CBA), Classification based on Multiple Association Rules (CMAR), and Fast Associative Classification Algorithm (FACA).