Thien Nguyen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Thien Nguyen has a total of 22 co-authors. Here are the top co-authors by number of shared publications:
- Phuoc Tran
Ton Duc Thang University
5 shared publications Huu Nguyen
Ho Chi Minh City University of Food Industry
2 shared publications- Huu Nguyen
Faculty of Information Technology, Ho Chi Minh City University of Food Industry, Ho Chi Minh City, Vietnam
2 shared publications - Huy Vo
Faculty of Information Technology, Ton Duc Thang University, Ho Chi Minh City, Vietnam
2 shared publications - Nguyen Le
Faculty of Information Technology, Ton Duc Thang University, Ho Chi Minh City, Vietnam
2 shared publications - T. T. Nguyen
Faculty of Information Technology, University of Science, Ho Chi Minh City, Vietnam
2 shared publications
Latest Publications
- Open Access
- 19 October 2026
- Published by Springer Nature in Journal of King Saud University Computer and Information Sciences
- Vol. 38 (9) , 929
Exact and heuristic approaches for mining top-k high-utility itemsets in uncertain databases with mixed utilities
- 19 October 2026
- Published by Springer Nature in Journal of King Saud University Computer and Information Sciences
- Vol. 38 (9) , 929
Open AccessExact and heuristic approaches for mining top-k high-utility itemsets in uncertain databases with mixed utilities
- Open Access
- 17 June 2026
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 21 (6) , e0351951
Best-first search–based approach for mining top-k closed frequent itemsets from uncertain databases
Uncertain data mining has become critical due to data generated by sensor networks, RFID systems, and data integration platforms. Mining top-k closed frequent itemsets from uncertain databases is particularly challenging because probabilistic support evaluation is expensive and the search space grows exponentially. Most existing methods rely on depth-first search (DFS) traversal, which explores candidates in enumeration order and often discovers high-support patterns late, leading to weak pruning and costly closure verification. This paper proposes TUFCI, a best-first-search-based algorithm for mining top-k closed frequent itemsets from uncertain databases. TUFCI explores candidates in descending order of probabilistic support using a priority queue, enabling early discovery of strong patterns, rapid threshold elevation, and safe early termination. Support-ordered exploration also improves closure checking by prioritizing supersets most likely to violate the closure property, thereby reducing redundant superset examinations. Experimental results demonstrate that TUFCI significantly outperforms DFS-based approaches in runtime and reduces the number of closure checks, especially on dense datasets.
- 17 June 2026
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 21 (6) , e0351951
Open AccessBest-first search–based approach for mining top-k closed frequent itemsets from uncertain databases
Uncertain data mining has become critical due to data generated by sensor networks, RFID systems, and data integration platforms. Mining top-k closed frequent itemsets from uncertain databases is particularly challenging because probabilistic support evaluation is expensive and the search space grows exponentially. Most existing methods rely on depth-first search (DFS) traversal, which explores candidates in enumeration order and often discovers high-support patterns late, leading to weak pruning and costly closure verification. This paper proposes TUFCI, a best-first-search-based algorithm for mining top-k closed frequent itemsets from uncertain databases. TUFCI explores candidates in descending order of probabilistic support using a priority queue, enabling early discovery of strong patterns, rapid threshold elevation, and safe early termination. Support-ordered exploration also improves closure checking by prioritizing supersets most likely to violate the closure property, thereby reducing redundant superset examinations. Experimental results demonstrate that TUFCI significantly outperforms DFS-based approaches in runtime and reduces the number of closure checks, especially on dense datasets.
- Open Access
- 18 November 2025
- Published by PeerJ in PeerJ Computer Science
- Vol. 11, e3304
Mining highly reliable dense subgraphs from uncertain weighted graphs for protein-protein interaction networks
Graphs are widely used as essential modeling tools to represent protein-protein interaction networks. However, they are often subject to uncertainty due to noise or incomplete data collection. This research aims to address the challenges associated with uncertain weighted graphs, where each edge is characterized by both a probability of existence and a numerical weight. We propose two novel greedy algorithms, GreedyUWDS and GreedyBWDS, designed to extract dense and reliable subgraphs from uncertain weighted graphs. GreedyUWDS optimizes subgraph density by leveraging the expected weighted density metric, whereas GreedyBWDS introduces a flexible reliability threshold parameter to effectively balance density and reliability, enabling users to adjust the algorithm to prioritize either highly reliable subgraphs at the expense of density or greater density at the expense of reliability, thereby addressing a critical trade-off that previous methods have not effectively managed. Experimental validation on protein-protein interaction networks demonstrates that the proposed algorithms outperform existing techniques in terms of weighted edge density and subgraph reliability. Specifically, GreedyBWDS achieves approximately a 20% improvement in graph density over GreedyUWDS, while reducing runtime by a factor of 2.5. GreedyBWDS consistently identifies subgraphs with higher values for the combined density and reliability objective, and demonstrates significantly greater computational efficiency. These results underscore the flexibility and efficiency of our proposed approaches, offering practical solutions for analyzing uncertain weighted graphs in biological contexts, with a primary focus on protein-protein interaction networks. This research advances the field of graph mining by enabling users to fine-tune the balance between density and reliability according to their specific application requirements.
- 18 November 2025
- Published by PeerJ in PeerJ Computer Science
- Vol. 11, e3304
Open AccessMining highly reliable dense subgraphs from uncertain weighted graphs for protein-protein interaction networks
Graphs are widely used as essential modeling tools to represent protein-protein interaction networks. However, they are often subject to uncertainty due to noise or incomplete data collection. This research aims to address the challenges associated with uncertain weighted graphs, where each edge is characterized by both a probability of existence and a numerical weight. We propose two novel greedy algorithms, GreedyUWDS and GreedyBWDS, designed to extract dense and reliable subgraphs from uncertain weighted graphs. GreedyUWDS optimizes subgraph density by leveraging the expected weighted density metric, whereas GreedyBWDS introduces a flexible reliability threshold parameter to effectively balance density and reliability, enabling users to adjust the algorithm to prioritize either highly reliable subgraphs at the expense of density or greater density at the expense of reliability, thereby addressing a critical trade-off that previous methods have not effectively managed. Experimental validation on protein-protein interaction networks demonstrates that the proposed algorithms outperform existing techniques in terms of weighted edge density and subgraph reliability. Specifically, GreedyBWDS achieves approximately a 20% improvement in graph density over GreedyUWDS, while reducing runtime by a factor of 2.5. GreedyBWDS consistently identifies subgraphs with higher values for the combined density and reliability objective, and demonstrates significantly greater computational efficiency. These results underscore the flexibility and efficiency of our proposed approaches, offering practical solutions for analyzing uncertain weighted graphs in biological contexts, with a primary focus on protein-protein interaction networks. This research advances the field of graph mining by enabling users to fine-tune the balance between density and reliability according to their specific application requirements.
- Open Access
- 19 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 165222-165246
Algorithms Based on Dynamic Minimum Probabilistic Support and Utility Thresholds for Mining Top-K High-Utility Itemsets From Uncertain Databases
Mining high-utility itemsets (HUIs) from large and uncertain databases is challenging due to the vast data volume and extensive search space. Determining the minimum utility and probabilistic support for all items is also a time-consuming task. Setting these thresholds too high results in too few itemsets, whereas setting them too low leads to an excessive number of itemsets and high computational costs. To address these issues, we developed solutions for mining Top-K high-utility itemsets from uncertain databases. First, we formally define the problem of mining Top-K high-utility itemsets from uncertain databases. Based on this formulation, we propose four algorithms: FTKHUUIM+, ITUHUFP, TUHUFP, and TKUU, which effectively meet the problem requirements. These algorithms adopt automated threshold-raising strategies and specialized storage structures to ensure optimal performance. Experimental results on public benchmark datasets demonstrate the scalability and robust performance of the proposed algorithms across both dense and sparse datasets in terms of runtime and efficiency. All of them share the common feature of pruning candidate itemsets based on dynamic minimum probabilistic support and utility thresholds. Although their pruning strategies differ, all algorithms follow a recursive process to select Top-K HUIs by pairing items from the database and filtering patterns according to the thresholds. Among the proposed algorithms, the ITUHUFP algorithm shows the best performance and stability on various benchmark datasets.
- 19 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 165222-165246
Open AccessAlgorithms Based on Dynamic Minimum Probabilistic Support and Utility Thresholds for Mining Top-K High-Utility Itemsets From Uncertain Databases
Mining high-utility itemsets (HUIs) from large and uncertain databases is challenging due to the vast data volume and extensive search space. Determining the minimum utility and probabilistic support for all items is also a time-consuming task. Setting these thresholds too high results in too few itemsets, whereas setting them too low leads to an excessive number of itemsets and high computational costs. To address these issues, we developed solutions for mining Top-K high-utility itemsets from uncertain databases. First, we formally define the problem of mining Top-K high-utility itemsets from uncertain databases. Based on this formulation, we propose four algorithms: FTKHUUIM+, ITUHUFP, TUHUFP, and TKUU, which effectively meet the problem requirements. These algorithms adopt automated threshold-raising strategies and specialized storage structures to ensure optimal performance. Experimental results on public benchmark datasets demonstrate the scalability and robust performance of the proposed algorithms across both dense and sparse datasets in terms of runtime and efficiency. All of them share the common feature of pruning candidate itemsets based on dynamic minimum probabilistic support and utility thresholds. Although their pruning strategies differ, all algorithms follow a recursive process to select Top-K HUIs by pairing items from the database and filtering patterns according to the thresholds. Among the proposed algorithms, the ITUHUFP algorithm shows the best performance and stability on various benchmark datasets.
- 12 July 2025
- Published by Springer Nature
- p. 308-321
Combination of Traditional Methods for Ukrainian-Vietnamese Deep Translation Models
- 12 July 2025
- Published by Springer Nature
- p. 308-321
Combination of Traditional Methods for Ukrainian-Vietnamese Deep Translation Models
- 12 July 2025
- Published by Springer Nature
- p. 234-247
Predicting the Likelihood of Discontinuing Banking Services Using Probabilistic Graphical Models
- 12 July 2025
- Published by Springer Nature
- p. 234-247
Predicting the Likelihood of Discontinuing Banking Services Using Probabilistic Graphical Models
- 12 July 2025
- Published by Springer Nature
- p. 230-240
Time Series Classification Using HMM and GMM with Pomegranate Library
- 12 July 2025
- Published by Springer Nature
- p. 230-240
Time Series Classification Using HMM and GMM with Pomegranate Library
- Open Access
- 10 March 2023
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Asian and Low-Resource Language Information Processing
- Vol. 22 (3) , 1-14
Building a Closed-Domain Question Answering System for a Low-Resource Language
In recent years, the Question Answering System (QAS) has been widely used to develop many systems, such as conversation systems, chatbots, and intelligent search. Depending on the amount of information or knowledge that the system processes, the system can be applied in answering the questions in an open domain or closed domain. There are many approaches to solving the QA problem, but the neural network models have yielded impressive and promising results, especially the Machine Reading Comprehension approach. In this article, we build a closed-domain QAS for a low-resource language, Vietnamese—specifically, “The Postgraduate Admission of Ho Chi Minh City University of Food Industry, Vietnam.” In addition, we have created two datasets to serve our QAS: vi-SQuAD v1.1, which is automatically translated and edited from SQuAD (Stanford University Question Answering Dataset), and HUFI-PostGrad, which is manually collected. We use two main models for the system, including the Intent Classification model and the Machine Reading Comprehension model. Experimental results initially show that our QAS gives encouraging results.
- 10 March 2023
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Asian and Low-Resource Language Information Processing
- Vol. 22 (3) , 1-14
Open AccessBuilding a Closed-Domain Question Answering System for a Low-Resource Language
In recent years, the Question Answering System (QAS) has been widely used to develop many systems, such as conversation systems, chatbots, and intelligent search. Depending on the amount of information or knowledge that the system processes, the system can be applied in answering the questions in an open domain or closed domain. There are many approaches to solving the QA problem, but the neural network models have yielded impressive and promising results, especially the Machine Reading Comprehension approach. In this article, we build a closed-domain QAS for a low-resource language, Vietnamese—specifically, “The Postgraduate Admission of Ho Chi Minh City University of Food Industry, Vietnam.” In addition, we have created two datasets to serve our QAS: vi-SQuAD v1.1, which is automatically translated and edited from SQuAD (Stanford University Question Answering Dataset), and HUFI-PostGrad, which is manually collected. We use two main models for the system, including the Intent Classification model and the Machine Reading Comprehension model. Experimental results initially show that our QAS gives encouraging results.
- Open Access
- 29 June 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 78928-78938
A Method of Chinese-Vietnamese Bilingual Corpus Construction for Machine Translation
A bilingual corpus is vital for natural language processing problems, especially in machine translation. The larger and better quality the corpus is, the higher the efficiency of the resulting machine translation is. There are two popular approaches to building a bilingual corpus. The first is building one automatically based on resources that are available on the internet, typically bilingual websites. The second approach is to construct one manually. Automated construction methods are being used more frequently because they are less expensive and there are a growing number of bilingual websites to exploit. In this paper, we use automated collection methods for a bilingual website to create a bilingual Chinese-Vietnamese corpus. In particular, the bilingual website we use to collect the data is the website of a multilingual dictionary (https://glosbe.com). We collected the Chinese-Vietnamese corpus from this website that includes more than 400k sentence pairs. We chose 100,000 sentence pairs in this corpus for machine translation experiments. From the corpus, we built five datasets consisting of 20k, 40k, 60k, 80k, and 100k sentence pairs, respectively. In addition, we built five additional datasets, applying word segmentation on the sentences of the original datasets. The experimental results showed that: (1) the quality of the corpus is relatively good with the highest BLEU score of 19.8, although there are still some issues that need to be addressed in future works; (2) the larger the corpus is, the higher the machine translation quality is; and (3) the untokenized datasets help train better translation models than the tokenized datasets.
- 29 June 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 78928-78938
Open AccessA Method of Chinese-Vietnamese Bilingual Corpus Construction for Machine Translation
A bilingual corpus is vital for natural language processing problems, especially in machine translation. The larger and better quality the corpus is, the higher the efficiency of the resulting machine translation is. There are two popular approaches to building a bilingual corpus. The first is building one automatically based on resources that are available on the internet, typically bilingual websites. The second approach is to construct one manually. Automated construction methods are being used more frequently because they are less expensive and there are a growing number of bilingual websites to exploit. In this paper, we use automated collection methods for a bilingual website to create a bilingual Chinese-Vietnamese corpus. In particular, the bilingual website we use to collect the data is the website of a multilingual dictionary (https://glosbe.com). We collected the Chinese-Vietnamese corpus from this website that includes more than 400k sentence pairs. We chose 100,000 sentence pairs in this corpus for machine translation experiments. From the corpus, we built five datasets consisting of 20k, 40k, 60k, 80k, and 100k sentence pairs, respectively. In addition, we built five additional datasets, applying word segmentation on the sentences of the original datasets. The experimental results showed that: (1) the quality of the corpus is relatively good with the highest BLEU score of 19.8, although there are still some issues that need to be addressed in future works; (2) the larger the corpus is, the higher the machine translation quality is; and (3) the untokenized datasets help train better translation models than the tokenized datasets.