Publications

@article{vijitkunsawat2023video,

title = {Video-Based Sign Language Digit Recognition for the Thai Language: A New Dataset and Method Comparisons},

author = {Wuttichai Vijitkunsawat and Teeradaj Racharak and Chau Nguyen and Nguyen Le Minh},

year  = {2023},

date = {2023-01-01},

urldate = {2023-01-01},

journal = {International Conference on Pattern Recognition Applications and Methods},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Nguyen2022,

title = {PhraseTransformer: an incorporation of local context information into sequence-to-sequence semantic parsing},

author = {Phuong Minh Nguyen and Tung Le and Huy Tien Nguyen and Vu Tran and Minh Le Nguyen},

url = {https://doi.org/10.1007/s10489-022-04246-0},

doi = {10.1007/s10489-022-04246-0},

issn = {1573-7497},

year  = {2022},

date = {2022-11-29},

urldate = {2022-11-29},

journal = {Applied Intelligence},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Viriyavisuthisakul2022,

title = {Parametric regularization loss in super-resolution reconstruction},

author = {Supatta Viriyavisuthisakul and Natsuda Kaothanthong and Parinya Sanguansat and Minh Le Nguyen and Choochart Haruechaiyasak},

url = {https://link.springer.com/article/10.1007/s00138-022-01315-9},

doi = {10.1007/S00138-022-01315-9/FIGURES/11},

issn = {14321769},

year  = {2022},

date = {2022-09-01},

urldate = {2022-09-01},

journal = {Machine Vision and Applications},

volume = {33},

number = {5},

pages = {1--21},

publisher = {Springer Science and Business Media Deutschland GmbH},

keywords = {Generative adversarial network, Image reconstruction, Loss function, Parametric, Regularization, Super-resolution},

pubstate = {published},

tppubtype = {article}

}

@article{viriyavisuthisakulregularization,

title = {A regularization-based generative adversarial network for single image super-resolution},

author = {Supatta Viriyavisuthisakul and Natsuda Kaothanthong and Parinya Sanguansat and Teeradaj Racharak and Minh Le Nguyen and Choochart Haruechaiyasak and Toshihiko Yamasaki},

url = {https://researchmap.jp/tracharak/published_papers/36200846/attachment_file.pdf},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{ye2022,

title = {Extractive Elementary Discourse Units for Improving Abstractive Summarization},

author = {Xiong Ye and Teeradaj Racharak and Minh Le Nguyen},

url = {/is/labs/nguyen-lab/home/wp-content/publications/Extractive%20Elementary%20Discourse%20Units%20for%20Improving%20Abstractive%20Summarization.pdf?_t=1653542221},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{kong_semanticweb,

title = {TransHExt: a Weighted Extension for TransH on Weighted Knowledge Graph Embedding},

author = {Wei Kun Kong and Xin Liu and Racharak Teeradaj and Minh Le Nguyen},

year  = {2022},

date = {2022-01-01},

journal = {The 21st International Semantic Web Conference},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{nguyen2022attentive,

title = {Attentive deep neural networks for legal document retrieval},

author = {Ha-Thanh Nguyen and Manh-Kien Phi and Xuan-Bach Ngo and Vu Tran and Le-Minh Nguyen and Minh-Phuong Tu},

year  = {2022},

date = {2022-01-01},

journal = {Artificial Intelligence and Law},

pages = {1--30},

publisher = {Springer},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{yuntao2022effective,

title = {An Effective Method to Answer Multi-hop Questions by Single-hop QA System},

author = {Kong Yuntao and Nguyen Minh Phuong and Teeradaj Racharak and Tung Le and Nguyen Le Minh},

url = {https://www.scitepress.org/Papers/2022/108242/108242.pdf},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{dang2022subtst,

title = {SubTST: A Combination of Sub-word Latent Topics and Sentence Transformer for Semantic Similarity Detection},

author = {Binh Dang and Tran-Thai Dang and Le-Minh Nguyen},

url = {https://www.scitepress.org/Papers/2022/107751/107751.pdf},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{thanh2022logical,

title = {Logical Structure-based Pretrained Models for Legal Text Processing},

author = {Nguyen Ha Thanh and Nguyen Le Minh},

url = {https://www.scitepress.org/Papers/2022/108520/108520.pdf},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{nguyen2022transformer,

title = {Transformer-Based Approaches for Legal Text Processing},

author = {Ha-Thanh Nguyen and Minh-Phuong Nguyen and Thi-Hai-Yen Vuong and Minh-Quan Bui and Minh-Chau Nguyen and Tran-Binh Dang and Vu Tran and Le-Minh Nguyen and Ken Satoh},

url = {https://link.springer.com/content/pdf/10.1007/s12626-022-00102-2.pdf},

year  = {2022},

date = {2022-01-01},

journal = {The Review of Socionetwork Strategies},

pages = {1--21},

publisher = {Springer Japan},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Nguyen2021113,

title = {Few-Shot Tuning Framework for Automated Terms of Service Generation},

author = {Ha Thanh Nguyen and Kiyoaki Shirai and Le Minh Nguyen},

url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85122085858&doi=10.3233%2fFAIA210325&partnerID=40&md5=433d5c2a792580ad3c830f5b9af17919},

doi = {10.3233/FAIA210325},

year  = {2021},

date = {2021-01-01},

urldate = {2021-01-01},

journal = {Frontiers in Artificial Intelligence and Applications},

volume = {346},

pages = {113-118},

abstract = {In this paper, we introduce BART2S a novel framework based on BART pretrained models to generate terms of service in high quality. The framework contains two parts: a generator finetuned with multiple tasks and a discriminator fine-tuned to distinguish the fair and unfair terms. Besides the novelty in design and the implementation contributions, the proposed framework can support drafting terms of service, a growing need in the digital age. Our proposed approach allows the system to reach a balance between automation and the will expression of the service provider. Through experiments, we demonstrate the effectiveness of the method and discuss potential future directions. © 2021 The authors and IOS Press.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{nguyen2021hydra,

title = {HYDRA--Hyper Dependency Representation Attentions},

author = {Ha-Thanh Nguyen and Vu Tran and Tran-Binh Dang and Minh-Quan Bui and Minh-Phuong Nguyen and Le-Minh Nguyen},

url = {https://arxiv.org/pdf/2109.05349.pdf},

year  = {2021},

date = {2021-01-01},

urldate = {2021-01-01},

journal = {ärXiv preprint arXiv:2109.05349},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{le2021multi,

title = {Multi visual and textual embedding on visual question answering for blind people},

author = {Tung Le and Huy Tien Nguyen and Minh Le Nguyen},

url = {https://www.sciencedirect.com/science/article/pii/S092523122101328X},

year  = {2021},

date = {2021-01-01},

urldate = {2021-01-01},

journal = {Neurocomputing},

volume = {465},

pages = {451--464},

publisher = {Elsevier},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Nguyen2021280,

title = {Knowledge Injection to Neural Networks with Progressive Learning Strategy},

author = {Ha Thanh Nguyen and Trung Kien Vu and Teeradaj Racharak and Le Minh Nguyen and Satoshi Tojo},

url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85103502197&doi=10.1007%2f978-3-030-71158-0_13&partnerID=40&md5=0ef8cbc30bed0f3b6b7d861c23360864},

doi = {10.1007/978-3-030-71158-0_13},

year  = {2021},

date = {2021-01-01},

journal = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)},

volume = {12613 LNAI},

pages = {280-290},

abstract = {Nowadays, deep learning has become the most modern and practical approach to solve a wide range of problems. With the ability to automatically extract the hierarchy of semantic level from the data, neural networks often outperform other techniques in complex issues. However, to perform well, the models need a vast amount of data, which is not always available. To overcome that problem, we propose an approach of injecting knowledge into the neural network instead of letting it struggles by itself. Our proposed policy for the training process is guiding the model to learn the label from a similarity distribution. Finally, we conduct experiments in the chord modeling problem to show the effectiveness of our method. © 2021, Springer Nature Switzerland AG.},

note = {cited By 0},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{nguyen2021paralaw,

title = {ParaLaw Nets--Cross-lingual Sentence-level Pretraining for Legal Text Processing},

author = {Ha-Thanh Nguyen and Vu Tran and Phuong Minh Nguyen and Thi-Hai-Yen Vuong and Quan Minh Bui and Chau Minh Nguyen and Binh Tran Dang and Minh Le Nguyen and Ken Satoh},

url = {https://arxiv.org/pdf/2106.13403.pdf},

year  = {2021},

date = {2021-01-01},

journal = {ärXiv preprint arXiv:2106.13403"},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{nguyen2021jnlp,

title = {Jnlp team: Deep learning approaches for legal processing tasks in coliee 2021},

author = {Ha-Thanh Nguyen and Phuong Minh Nguyen and Thi-Hai-Yen Vuong and Quan Minh Bui and Chau Minh Nguyen and Binh Tran Dang and Vu Tran and Minh Le Nguyen and Ken Satoh},

url = {https://arxiv.org/pdf/2106.13405.pdf},

year  = {2021},

date = {2021-01-01},

journal = {ärXiv preprint arXiv:2106.13405"},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{nguyen2021sublanguage,

title = {Sublanguage: A serious issue affects pretrained models in legal domain},

author = {Ha-Thanh Nguyen and Le-Minh Nguyen},

url = {https://arxiv.org/pdf/2104.07782.pdf},

year  = {2021},

date = {2021-01-01},

journal = {ärXiv preprint arXiv:2104.07782"},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{nguyen2021scnn,

title = {SCNN: Swarm Characteristic Neural Network},

author = {Ha-Thanh Nguyen and Le-Minh Nguyen},

url = {https://arxiv.org/pdf/2103.15550.pdf},

year  = {2021},

date = {2021-01-01},

urldate = {2021-01-01},

journal = {ärXiv preprint arXiv:2103.15550},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Tran202144,

title = {ReLink: Open information extraction by linking phrases and its applications},

author = {Xuan-Chien Tran and Le-Minh Nguyen},

url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85098284935&doi=10.1007%2f978-3-030-65621-8_3&partnerID=40&md5=dc46a23c029804cc93c94547951d04a5},

doi = {10.1007/978-3-030-65621-8_3},

year  = {2021},

date = {2021-01-01},

urldate = {2021-01-01},

journal = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)},

volume = {12582 LNCS},

pages = {44-62},

abstract = {Recently, many Open IE systems have been developed based on using deep linguistic features such as dependency-parse features to overcome the limitations presented in older Open IE systems which use only shallow information like part-of-speech or chunking. Even though these newer systems have some clear advantages in their extractions, they also possess several issues which do not exist in old systems. In this paper, we analyze the outputs from several popular Open IE systems to find out their strength and weaknesses. Then we introduce ReLink, a novel Open IE system for extracting binary relations from open-domain text. Its working model is based on identifying correct phrases and linking them in the most proper way to reflect their relationship in a sentence. After establishing connections, it can easily extract relations by using several pre-defined patterns. Despite using only shallow linguistic features for extraction, it does not have the same weakness that existed in older systems, and it can also avoid many similar issues arising in recent Open IE systems. Our experiments show that ReLink achieves larger Area Under Precision-Recall Curve compared with ReVerb and Ollie, two well-known Open IE systems. © Springer Nature Switzerland AG 2021.},

note = {cited By 0},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Tran2021,

title = {Variational model for low-resource natural language generation in spoken dialogue systems},

author = {Van-Khanh Tran and Le-Minh Nguyen},

url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85086829108&doi=10.1016%2fj.csl.2020.101120&partnerID=40&md5=35f5627aa2f410d78eaf03bddde87b74},

doi = {10.1016/j.csl.2020.101120},

year  = {2021},

date = {2021-01-01},

urldate = {2021-01-01},

journal = {Computer Speech and Language},

volume = {65},

abstract = {Natural Language Generation (NLG) plays a critical role in Spoken Dialogue Systems (SDSs), aims at converting a meaning representation into natural language utterances. Recent deep learning-based generators have shown improving results irrespective of providing sufficient annotated data. Nevertheless, how to build a generator that can effectively utilize as much of knowledge from a low-resource setting data is a crucial issue for NLG in SDSs. This paper presents a variational-based NLG framework to tackle the NLG problem of having limited annotated data in two scenarios, domain adaptation and low-resource in-domain training data. Based on this framework, we propose a novel adversarial domain adaptation NLG taclking the former issue, while the latter issue is also handled by a second proposed dual variational model. We extensively conducted the experiments on four different domains in a variety of training scenarios, in which the experimental results show that the proposed methods not only outperform previous methods when having sufficient training dataset but also show its ability to work acceptably well when there is a small amount of in-domain data or adapt quickly to a new domain with only a low-resource target domain data. © 2020},

note = {cited By 2},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Nguyen2020497,

title = {Deep Learning Approach for Vietnamese Consonant Misspell Correction},

author = {Ha Thanh Nguyen and Tran Binh Dang and Le Minh Nguyen},

url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85088519418&doi=10.1007%2f978-981-15-6168-9_40&partnerID=40&md5=6d477730cc2c02551bd7fac24d188c48},

doi = {10.1007/978-981-15-6168-9_40},

year  = {2020},

date = {2020-01-01},

journal = {Communications in Computer and Information Science},

volume = {1215 CCIS},

pages = {497-504},

abstract = {Vietnamese words are combinations of consonants, vowels, and diacritics. Previous studies on Vietnamese spelling correction often focused on mistyped errors. Misspelled errors are more common and difficult to detect. Based on our literature review, there is no direct study to address this issue. A misspelled Vietnamese word can become another word does exist in the vocabulary but make the sentence a different meaning or meaningless. While mistyped errors are typographical errors, misspelled errors may appear in any type of text including typed documents and handwritten text. Compared to mistyped errors, misspelled errors are harder to detect, especially by people who type it out. This error comes from the wrong understanding about the spelling of the word. For that reason, checking a sentence with a vocabulary filter does not guarantee that the sentence is spelled correctly. Checking Vietnamese spelling errors is a difficult problem. There have been many articles trying to solve this problem with different approaches but they have their own limitations. In this paper, we propose a deep learning approach focusing on consonant misspell errors with superior accuracy compared to the existing methods. The accuracy of our model makes a significant gap compared to the current state-of-the-art model. © 2020, Springer Nature Singapore Pte Ltd.},

note = {cited By 0},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{nguyen2020jnlp,

title = {Jnlp team: Deep learning for legal processing in coliee 2020},

author = {Ha-Thanh Nguyen and Hai-Yen Thi Vuong and Phuong Minh Nguyen and Binh Tran Dang and Quan Minh Bui and Sinh Trong Vu and Chau Minh Nguyen and Vu Tran and Ken Satoh and Minh Le Nguyen},

url = {https://arxiv.org/pdf/2011.08071.pdf},

year  = {2020},

date = {2020-01-01},

journal = {ärXiv preprint arXiv:2011.08071"},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{tran2020encoded,

title = {Encoded summarization: summarizing documents into continuous vector space for legal case retrieval},

author = {Vu Tran and Minh Le Nguyen and Satoshi Tojo and Ken Satoh},

url = {https://link.springer.com/article/10.1007/s10506-020-09262-4},

year  = {2020},

date = {2020-01-01},

urldate = {2020-01-01},

journal = {Ärtificial Intelligence and Law},

volume = {28},

number = {4},

pages = {441--467},

publisher = {Springer Netherlands},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Nguyen2020,

title = {ILWAANet: An Interactive Lexicon-Aware Word-Aspect Attention Network for aspect-level sentiment classification on social networking},

author = {Huy-Thanh Nguyen and Le-Minh Nguyen},

url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85077942279&doi=10.1016%2fj.eswa.2019.113065&partnerID=40&md5=bb19ebc7b9d3cdf19543e9a65cd5c919},

doi = {10.1016/j.eswa.2019.113065},

year  = {2020},

date = {2020-01-01},

urldate = {2020-01-01},

journal = {Expert Systems with Applications},

volume = {146},

abstract = {Än Interactive Lexicon-Aware Word-Aspect Attention Network (ILWAAN) is proposed for aspect-level sentiment classification which deals with identifying the sentiment polarity of a specific aspect in its context and have potential application on social networking. In this model, effective multiple attention mechanisms (intra-attention and interactive-attention mechanisms) integrated with sentiment lexicon information are developed to form an aspect-specific representation at two levels: Phrase-level and Aggregation-level information. Specifically, an aspect and its context are fused with the sentiment lexicon information and learn their relationship representations by lexicon-aware attention operations. This allows the model to tries to incorporate the aspect information into the deep neural networks and learn to attend the correct sentiment context words conditioned on the informative aspect words. To evaluate the performance, we evaluate our model in three benchmark data: Twitter, Laptop, and Restaurant. The experimental results indicate that our models improve the performance for aspect-level sentiment classification. © 2019"},

note = {cited By 12},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{ngo2019transformer,

title = {How Transformer Revitalizes Character-based Neural Machine Translation: An Investigation on Japanese-Vietnamese Translation Systems},

author = {Thi-Vinh Ngo and Thanh-Le Ha and Phuong-Thai Nguyen and Le-Minh Nguyen},

year  = {2019},

date = {2019-01-01},

journal = {ärXiv preprint arXiv:1910.02238"},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Tien2019,

title = {Sentence modeling via multiple word embeddings and multi-level comparison for semantic textual similarity},

author = {Nguyen Huy Tien and Nguyen Minh Le and Yamasaki Tomohiro and Izuha Tatsuya},

url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85070327866&doi=10.1016%2fj.ipm.2019.102090&partnerID=40&md5=fc4974c39bad804efd063dc9605b49b4},

doi = {10.1016/j.ipm.2019.102090},

year  = {2019},

date = {2019-01-01},

journal = {Information Processing and Management},

volume = {56},

number = {6},

abstract = {Recently, using a pretrained word embedding to represent words achieves success in many natural language processing tasks. According to objective functions, different word embedding models capture different aspects of linguistic properties. However, the Semantic Textual Similarity task, which evaluates similarity/relation between two sentences, requires to take into account of these linguistic aspects. Therefore, this research aims to encode various characteristics from multiple sets of word embeddings into one embedding and then learn similarity/relation between sentences via this novel embedding. Representing each word by multiple word embeddings, the proposed MaxLSTM-CNN encoder generates a novel sentence embedding. We then learn the similarity/relation between our sentence embeddings via Multi-level comparison. Our method M-MaxLSTM-CNN consistently shows strong performances in several tasks (i.e., measure textual similarity, identify paraphrase, recognize textual entailment). Our model does not use hand-crafted features (e.g., alignment features, Ngram overlaps, dependency features) as well as does not require pre-trained word embeddings to have the same dimension. © 2019 Elsevier Ltd},

note = {cited By 24},

keywords = {},

pubstate = {published},

tppubtype = {article}

}