2019
140.
Sinh, Vu Trong; Minh, Nguyen Le
A Study on Self-attention Mechanism for AMR-to-text Generation Journal Article
In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11608 LNCS, pp. 321-328, 2019, (cited By 0).
Abstract | Links | BibTeX | Tags:
@article{Sinh2019321,
title = {A Study on Self-attention Mechanism for AMR-to-text Generation},
author = {Vu Trong Sinh and Nguyen Le Minh},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85068326115&doi=10.1007%2f978-3-030-23281-8_27&partnerID=40&md5=099dc2e96c585494a492a5d3c3dca829},
doi = {10.1007/978-3-030-23281-8_27},
year = {2019},
date = {2019-01-01},
journal = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)},
volume = {11608 LNCS},
pages = {321-328},
abstract = {Introduced by Vaswani et al., transformer architecture, with the effective use of self-attention mechanism, has shown outstanding performance in translating sequence of text from one language to another. In this paper, we conduct experiments using the self-attention in converting an abstract meaning representation (AMR) graph, a semantic representation, into a natural language sentence, also known as the task of AMR-to-text generation. On the benchmark dataset for this task, we obtain promising results comparing to existing deep learning methods in the literature. © 2019, Springer Nature Switzerland AG.},
note = {cited By 0},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Introduced by Vaswani et al., transformer architecture, with the effective use of self-attention mechanism, has shown outstanding performance in translating sequence of text from one language to another. In this paper, we conduct experiments using the self-attention in converting an abstract meaning representation (AMR) graph, a semantic representation, into a natural language sentence, also known as the task of AMR-to-text generation. On the benchmark dataset for this task, we obtain promising results comparing to existing deep learning methods in the literature. © 2019, Springer Nature Switzerland AG.
139.
Nguyen, Minh-Tien; Tran, Viet Cuong; Nguyen, Xuan Hoai; Nguyen, Le-Minh
Web document summarization by exploiting social context with matrix co-factorization Journal Article
In: Information Processing and Management, vol. 56, no. 3, pp. 495-515, 2019, (cited By 14).
Abstract | Links | BibTeX | Tags:
@article{Nguyen2019495,
title = {Web document summarization by exploiting social context with matrix co-factorization},
author = {Minh-Tien Nguyen and Viet Cuong Tran and Xuan Hoai Nguyen and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85059551583&doi=10.1016%2fj.ipm.2018.12.006&partnerID=40&md5=65cb397e22003733491d825ae47b8f68},
doi = {10.1016/j.ipm.2018.12.006},
year = {2019},
date = {2019-01-01},
journal = {Information Processing and Management},
volume = {56},
number = {3},
pages = {495-515},
abstract = {In the context of social media, users usually post relevant information corresponding to the contents of events mentioned in a Web document. This information posses two important values in that (i) it reflects the content of an event and (ii) it shares hidden topics with sentences in the main document. In this paper, we present a novel model to capture the nature of relationships between document sentences and post information (comments or tweets) in sharing hidden topics for summarization of Web documents by utilizing relevant post information. Unlike previous methods which are usually based on hand-crafted features, our approach ranks document sentences and user posts based on their importance to the topics. The sentence-user-post relation is formulated in a share topic matrix, which presents their mutual reinforcement support. Our proposed matrix co-factorization algorithm computes the score of each document sentence and user post and extracts the top ranked document sentences and comments (or tweets) as a summary. We apply the model to the task of summarization on three datasets in two languages, English and Vietnamese, of social context summarization and also on DUC 2004 (a standard corpus of the traditional summarization task). According to the experimental results, our model significantly outperforms the basic matrix factorization and achieves competitive ROUGE-scores with state-of-the-art methods. © 2018 Elsevier Ltd},
note = {cited By 14},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
In the context of social media, users usually post relevant information corresponding to the contents of events mentioned in a Web document. This information posses two important values in that (i) it reflects the content of an event and (ii) it shares hidden topics with sentences in the main document. In this paper, we present a novel model to capture the nature of relationships between document sentences and post information (comments or tweets) in sharing hidden topics for summarization of Web documents by utilizing relevant post information. Unlike previous methods which are usually based on hand-crafted features, our approach ranks document sentences and user posts based on their importance to the topics. The sentence-user-post relation is formulated in a share topic matrix, which presents their mutual reinforcement support. Our proposed matrix co-factorization algorithm computes the score of each document sentence and user post and extracts the top ranked document sentences and comments (or tweets) as a summary. We apply the model to the task of summarization on three datasets in two languages, English and Vietnamese, of social context summarization and also on DUC 2004 (a standard corpus of the traditional summarization task). According to the experimental results, our model significantly outperforms the basic matrix factorization and achieves competitive ROUGE-scores with state-of-the-art methods. © 2018 Elsevier Ltd
138.
Nguyen, Van-Nhat; Nguyen, Ha-Thanh; Vo, Dinh-Hieu; Nguyen, Le-Minh
2019, (cited By 0).
Abstract | Links | BibTeX | Tags:
@conference{Nguyen201999,
title = {Relation Extraction in Vietnamese Text via Piecewise Convolution Neural Network with Word-Level Attention},
author = {Van-Nhat Nguyen and Ha-Thanh Nguyen and Dinh-Hieu Vo and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85061922441&doi=10.1109%2fNICS.2018.8606824&partnerID=40&md5=667afcb4c73744ba20dc5d56012c5bb1},
doi = {10.1109/NICS.2018.8606824},
year = {2019},
date = {2019-01-01},
journal = {NICS 2018 - Proceedings of 2018 5th NAFOSTED Conference on Information and Computer Science},
pages = {99-103},
abstract = {With the explosion of information technology, the Internet now contains enormous amounts of data, so the role of information extraction systems becomes very important. Relation Extraction is a sub-task of Information Extraction, which focuses on classifying the relationship between the entity pairs mentioned in the text. In recent years, despite the many new methods have been introduced, Relation Extraction still receives attention from researchers for languages in general and Vietnamese in particular.Relation Extraction can be addressed in a variety of ways, including supervised learning methods, unsupervised and semi-supervised methods. Recent studies in the English language have shown that Relation Extraction using deep learning method in the supervised or semi-supervised domains is achieving optimal and superior results over traditional non-deep learning methods. However, researches in Vietnamese are few and in the process of searching documents, the results of deep learning applying for Relation Extraction in Vietnamese are not found. Therefore, the research focuses on studying and research the method of using deep learning to solve Relation Extraction task in Vietnamese. In order to solve the Relation Extraction task, the research proposes and constructs a deep learning model named Piecewise Convolution Neural Network with Word-Level Attention. © 2018 IEEE.},
note = {cited By 0},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
With the explosion of information technology, the Internet now contains enormous amounts of data, so the role of information extraction systems becomes very important. Relation Extraction is a sub-task of Information Extraction, which focuses on classifying the relationship between the entity pairs mentioned in the text. In recent years, despite the many new methods have been introduced, Relation Extraction still receives attention from researchers for languages in general and Vietnamese in particular.Relation Extraction can be addressed in a variety of ways, including supervised learning methods, unsupervised and semi-supervised methods. Recent studies in the English language have shown that Relation Extraction using deep learning method in the supervised or semi-supervised domains is achieving optimal and superior results over traditional non-deep learning methods. However, researches in Vietnamese are few and in the process of searching documents, the results of deep learning applying for Relation Extraction in Vietnamese are not found. Therefore, the research focuses on studying and research the method of using deep learning to solve Relation Extraction task in Vietnamese. In order to solve the Relation Extraction task, the research proposes and constructs a deep learning model named Piecewise Convolution Neural Network with Word-Level Attention. © 2018 IEEE.
137.
Nguyen, Minh-Tien; Lai, Dac Viet; Nguyen, Huy Tien; Nguyen, Minh Le
Tsix: A Human-involved-creation Dataset for Tweet Summarization Conference
2019, (cited By 3).
Abstract | Links | BibTeX | Tags:
@conference{Nguyen20193204,
title = {Tsix: A Human-involved-creation Dataset for Tweet Summarization},
author = {Minh-Tien Nguyen and Dac Viet Lai and Huy Tien Nguyen and Minh Le Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85059904115&partnerID=40&md5=4f12f4bd88889e99948ce9797be2d648},
year = {2019},
date = {2019-01-01},
journal = {LREC 2018 - 11th International Conference on Language Resources and Evaluation},
pages = {3204-3208},
abstract = {We present a new dataset for tweet summarization. The dataset includes six events collected from Twitter from October 10 to November 9, 2016. Our dataset features two prominent properties. Firstly, human-annotated gold-standard references allow to correctly evaluate extractive summarization methods. Secondly, tweets are assigned into sub-topics divided by consecutive days, which facilitate incremental tweet stream summarization methods. To reveal the potential usefulness of our dataset, we compare several well-known summarization methods. Experimental results indicate that among extractive approaches, hybrid term frequency - document term frequency obtains competitive results in term of ROUGE-scores. The analysis also shows that polarity is an implicit factor of tweets in our dataset, suggesting that it can be exploited as a component besides tweet content quality in the summarization process. © LREC 2018 - 11th International Conference on Language Resources and Evaluation. All rights reserved.},
note = {cited By 3},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
We present a new dataset for tweet summarization. The dataset includes six events collected from Twitter from October 10 to November 9, 2016. Our dataset features two prominent properties. Firstly, human-annotated gold-standard references allow to correctly evaluate extractive summarization methods. Secondly, tweets are assigned into sub-topics divided by consecutive days, which facilitate incremental tweet stream summarization methods. To reveal the potential usefulness of our dataset, we compare several well-known summarization methods. Experimental results indicate that among extractive approaches, hybrid term frequency - document term frequency obtains competitive results in term of ROUGE-scores. The analysis also shows that polarity is an implicit factor of tweets in our dataset, suggesting that it can be exploited as a component besides tweet content quality in the summarization process. © LREC 2018 - 11th International Conference on Language Resources and Evaluation. All rights reserved.
136.
Tran, Van-Khanh; Nguyen, Le-Minh
Gating mechanism based Natural Language Generation for spoken dialogue systems Journal Article
In: Neurocomputing, vol. 325, pp. 48-58, 2019, (cited By 4).
Abstract | Links | BibTeX | Tags:
@article{Tran201948,
title = {Gating mechanism based Natural Language Generation for spoken dialogue systems},
author = {Van-Khanh Tran and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85055287829&doi=10.1016%2fj.neucom.2018.09.069&partnerID=40&md5=f945e3a0efa1a9db6d122fafc7145c56},
doi = {10.1016/j.neucom.2018.09.069},
year = {2019},
date = {2019-01-01},
journal = {Neurocomputing},
volume = {325},
pages = {48-58},
abstract = {Recurrent Neural Network (RNN) based approaches have recently shown promising in tackling Natural Language Generation (NLG) problems. This paper presents an approach to leverage gating mechanisms, in which we incrementally propose three additional semantic cells into a traditional RNN model: a Refinement cell to filter the sequential inputs before RNN computations, an Adjustment cell, and an Output cell to select semantic elements and gate a feature vector during generation. The proposed gating-based generators can learn from unaligned data by jointly training both sentence planning and surface realization to generate natural language utterances. We conducted extensive experiments on four different NLG domains in which the results empirically show that the proposed methods not only achieved better performance on all the NLG domains in comparison with previous gating-based, attention-based methods, but also obtained highly competitive results compared to a hybrid generator. © 2018 Elsevier B.V.},
note = {cited By 4},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Recurrent Neural Network (RNN) based approaches have recently shown promising in tackling Natural Language Generation (NLG) problems. This paper presents an approach to leverage gating mechanisms, in which we incrementally propose three additional semantic cells into a traditional RNN model: a Refinement cell to filter the sequential inputs before RNN computations, an Adjustment cell, and an Output cell to select semantic elements and gate a feature vector during generation. The proposed gating-based generators can learn from unaligned data by jointly training both sentence planning and surface realization to generate natural language utterances. We conducted extensive experiments on four different NLG domains in which the results empirically show that the proposed methods not only achieved better performance on all the NLG domains in comparison with previous gating-based, attention-based methods, but also obtained highly competitive results compared to a hybrid generator. © 2018 Elsevier B.V.
135.
Trieu, Hai-Long; Tran, Duc-Vu; Ittoo, Ashwin; Nguyen, Le-Minh
Leveraging additional resources for improving statistical machine translation on asian low-resource languages Journal Article
In: ÄCM Transactions on Asian and Low-Resource Language Information Processing", vol. 18, no. 3, 2019, (cited By 2).
Abstract | Links | BibTeX | Tags:
@article{Trieu2019,
title = {Leveraging additional resources for improving statistical machine translation on asian low-resource languages},
author = {Hai-Long Trieu and Duc-Vu Tran and Ashwin Ittoo and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85075758647&doi=10.1145%2f3314936&partnerID=40&md5=aace4f12b003e0135cdd783a3019482e},
doi = {10.1145/3314936},
year = {2019},
date = {2019-01-01},
journal = {ÄCM Transactions on Asian and Low-Resource Language Information Processing"},
volume = {18},
number = {3},
abstract = {Phrase-based machine translation (MT) systems require large bilingual corpora for training. Nevertheless, such large bilingual corpora are unavailable for most language pairs in the world, causing a bottleneck for the development of MT. For the Asian language pairs-Japanese, Indonesian, Malay paired with Vietnamese-they are also not excluded from the case, in which there are no large bilingual corpora on these low-resource language pairs. Furthermore, although the languages are widely used in the world, there is no prior work on MT, which causes an issue for the development of MT on these languages. In this article, we conducted an empirical study of leveraging additional resources to improve MT for the Asian low-resource language pairs: Translation fromJapanese, Indonesian, andMalay to Vietnamese.We propose an innovative approach that lies in two strategies of building bilingual corpora from comparable data and phrase pivot translation on existing bilingual corpora of the languages paired with English. Bilingual corpora were built from Wikipedia bilingual titles to enhance bilingual data for the low-resource languages. Additionally,we introduced a combined model of the additional resources to create an effective solution to improveMT on the Asian low-resource languages. Experimental results show the effectiveness of our systems with the improvement of +2 to +7 BLEU points. This work contributes to the development of MT on low-resource languages, especially opening a promising direction for the progress of MT on the Asian language pairs. © 2019 Association for Computing Machinery.},
note = {cited By 2},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Phrase-based machine translation (MT) systems require large bilingual corpora for training. Nevertheless, such large bilingual corpora are unavailable for most language pairs in the world, causing a bottleneck for the development of MT. For the Asian language pairs-Japanese, Indonesian, Malay paired with Vietnamese-they are also not excluded from the case, in which there are no large bilingual corpora on these low-resource language pairs. Furthermore, although the languages are widely used in the world, there is no prior work on MT, which causes an issue for the development of MT on these languages. In this article, we conducted an empirical study of leveraging additional resources to improve MT for the Asian low-resource language pairs: Translation fromJapanese, Indonesian, andMalay to Vietnamese.We propose an innovative approach that lies in two strategies of building bilingual corpora from comparable data and phrase pivot translation on existing bilingual corpora of the languages paired with English. Bilingual corpora were built from Wikipedia bilingual titles to enhance bilingual data for the low-resource languages. Additionally,we introduced a combined model of the additional resources to create an effective solution to improveMT on the Asian low-resource languages. Experimental results show the effectiveness of our systems with the improvement of +2 to +7 BLEU points. This work contributes to the development of MT on low-resource languages, especially opening a promising direction for the progress of MT on the Asian language pairs. © 2019 Association for Computing Machinery.
134.
Nguyen, Huy Tien; Nguyen, Minh Le
An ensemble method with sentiment features and clustering support Journal Article
In: Neurocomputing, vol. 370, pp. 155-165, 2019, (cited By 14).
Abstract | Links | BibTeX | Tags:
@article{Nguyen2019155,
title = {An ensemble method with sentiment features and clustering support},
author = {Huy Tien Nguyen and Minh Le Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85071983415&doi=10.1016%2fj.neucom.2019.08.071&partnerID=40&md5=cd9de9ba61553dfaeb98217e8d75b8fd},
doi = {10.1016/j.neucom.2019.08.071},
year = {2019},
date = {2019-01-01},
journal = {Neurocomputing},
volume = {370},
pages = {155-165},
abstract = {Long Short Term Memory (LSTM) and Convolutional Neural Network (CNN) are efficiently applied to natural language processing, especially sentiment analysis. CNN employs filters to capture local dependencies while LSTM designs a cell to memorize long-distance information. However, integrating these advantages into one model is challenging because of overfitting in training. To avoid this problem, we propose a freezing technique to learn sentiment-specific vectors from CNN and LSTM. This technique is efficient for integrating the advantages of various deep learning models. We also observe that semantically clustering documents into groups is more beneficial for ensemble methods. According to the experiments, our method achieves competitive results on the five well-known datasets: Pang & Lee movie reviews, Stanford Twitter Sentiment and Stanford Sentiment Treebank for sentence level, IMDB large movie reviews, and SenTube for document level. © 2019},
note = {cited By 14},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Long Short Term Memory (LSTM) and Convolutional Neural Network (CNN) are efficiently applied to natural language processing, especially sentiment analysis. CNN employs filters to capture local dependencies while LSTM designs a cell to memorize long-distance information. However, integrating these advantages into one model is challenging because of overfitting in training. To avoid this problem, we propose a freezing technique to learn sentiment-specific vectors from CNN and LSTM. This technique is efficient for integrating the advantages of various deep learning models. We also observe that semantically clustering documents into groups is more beneficial for ensemble methods. According to the experiments, our method achieves competitive results on the five well-known datasets: Pang & Lee movie reviews, Stanford Twitter Sentiment and Stanford Sentiment Treebank for sentence level, IMDB large movie reviews, and SenTube for document level. © 2019
133.
Trieu, Hai Long; Tran, Duc-Vu; Nguyen, Minh Le
Investigating phrase-based and neural-based machine translation on low-resource settings Conference
2019, (cited By 2).
Abstract | Links | BibTeX | Tags:
@conference{Trieu2019384,
title = {Investigating phrase-based and neural-based machine translation on low-resource settings},
author = {Hai Long Trieu and Duc-Vu Tran and Minh Le Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85072811470&partnerID=40&md5=5696f540c15611279faa6fa7af992eb9},
year = {2019},
date = {2019-01-01},
journal = {PACLIC 2017 - Proceedings of the 31st Pacific Asia Conference on Language, Information and Computation},
pages = {384-391},
abstract = {Neural-based and phrase-based methods have shown the effectiveness and promising results in the development of current machine translation. The two methods are compared on some European languages, which show the advantages of the neural machine translation. Nevertheless, there are few work of comparing the two methods on low-resource languages, which there are only small bilingual corpora. The problem of unavailable large bilingual corpora causes a bottleneck for machine translation for such language pairs. In this paper, we present a comparison of the phrase-based and neural-based machine translation methods on several Asian language pairs: Japanese-English, Indonesian-Vietnamese, and English-Vietnamese. Additionally, we extracted a bilingual corpus from Wikipedia to enhance machine translation performance. Experimental results showed that when using the extracted corpus to enlarge the training data, neural machine translation models achieved the higher improvement and outperformed the phrase-based models. This work can be useful as a basis for further development of machine translation on the low-resource languages. Copyright © 2017 Hai Long Trieu, Vu Tran and Nguyen Le Minh},
note = {cited By 2},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Neural-based and phrase-based methods have shown the effectiveness and promising results in the development of current machine translation. The two methods are compared on some European languages, which show the advantages of the neural machine translation. Nevertheless, there are few work of comparing the two methods on low-resource languages, which there are only small bilingual corpora. The problem of unavailable large bilingual corpora causes a bottleneck for machine translation for such language pairs. In this paper, we present a comparison of the phrase-based and neural-based machine translation methods on several Asian language pairs: Japanese-English, Indonesian-Vietnamese, and English-Vietnamese. Additionally, we extracted a bilingual corpus from Wikipedia to enhance machine translation performance. Experimental results showed that when using the extracted corpus to enlarge the training data, neural machine translation models achieved the higher improvement and outperformed the phrase-based models. This work can be useful as a basis for further development of machine translation on the low-resource languages. Copyright © 2017 Hai Long Trieu, Vu Tran and Nguyen Le Minh
132.
Vu, Trong Sinh; Nguyen, Le Minh
An Empirical Evaluation of AMR Parsing for Legal Documents Journal Article
In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11717 LNAI, pp. 131-145, 2019, (cited By 0).
Abstract | Links | BibTeX | Tags:
@article{Vu2019131,
title = {An Empirical Evaluation of AMR Parsing for Legal Documents},
author = {Trong Sinh Vu and Le Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85075556833&doi=10.1007%2f978-3-030-31605-1_11&partnerID=40&md5=43113718dc9930d7c0fa59501b957825},
doi = {10.1007/978-3-030-31605-1_11},
year = {2019},
date = {2019-01-01},
journal = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)},
volume = {11717 LNAI},
pages = {131-145},
abstract = {Many approaches have been proposed to tackle the problem of Abstract Meaning Representation (AMR) parsing, help solving various natural language processing issues recently. In our paper, we provide an overview of different methods in AMR parsing and their performances when analyzing legal documents. We conduct experiments of different AMR parsers on our annotated dataset extracted from the English version of Japanese Civil Code. Our results show the limitations as well as open a room for improvements of current parsing techniques when applying in this non-trivial domain. © 2019, Springer Nature Switzerland AG.},
note = {cited By 0},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Many approaches have been proposed to tackle the problem of Abstract Meaning Representation (AMR) parsing, help solving various natural language processing issues recently. In our paper, we provide an overview of different methods in AMR parsing and their performances when analyzing legal documents. We conduct experiments of different AMR parsers on our annotated dataset extracted from the English version of Japanese Civil Code. Our results show the limitations as well as open a room for improvements of current parsing techniques when applying in this non-trivial domain. © 2019, Springer Nature Switzerland AG.
131.
Nguyen, Huy Xuan; Nguyen, Le Minh
Attention mechanism for recommender systems Conference
2019, (cited By 0).
Abstract | Links | BibTeX | Tags:
@conference{Xuan2019174,
title = {Attention mechanism for recommender systems},
author = {Huy Xuan Nguyen and Le Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85084953230&partnerID=40&md5=0001269121e6afeb17c2ae3eb89a861d},
year = {2019},
date = {2019-01-01},
journal = {Proceedings of the 33rd Pacific Asia Conference on Language, Information and Computation, PACLIC 2019},
pages = {174-181},
abstract = {Sparseness of user item rating data affects the quality of recommender systems. To solve this problem, many approaches have been proposed. They added supplemental information to increase the accuracy. We propose a recommendation model namely attention matrix factorization (AMF) that integrates attention mechanism of the both item reviews document and item genre information into probabilistic matrix factorization (PMF). Consequently, AMF attends features which are mentioned in item reviews document and further increases the rating prediction accuracy by adding item genre information. Our experiments on the Movielens and Amazon instant video datasets show that AMF outperforms the previous traditional recommendation systems. This reveals that our model can capture subtle features of item reviews although the rating data is sparse. Copyright © 2019 Xuan-Huy Nguyen and Le-Minh Nguyen.},
note = {cited By 0},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Sparseness of user item rating data affects the quality of recommender systems. To solve this problem, many approaches have been proposed. They added supplemental information to increase the accuracy. We propose a recommendation model namely attention matrix factorization (AMF) that integrates attention mechanism of the both item reviews document and item genre information into probabilistic matrix factorization (PMF). Consequently, AMF attends features which are mentioned in item reviews document and further increases the rating prediction accuracy by adding item genre information. Our experiments on the Movielens and Amazon instant video datasets show that AMF outperforms the previous traditional recommendation systems. This reveals that our model can capture subtle features of item reviews although the rating data is sparse. Copyright © 2019 Xuan-Huy Nguyen and Le-Minh Nguyen.
130.
Vu, Sinh Trong; Nguyen, Minh Le; Satoh, Ken
Legal Text Generation from Abstract Meaning Representation Inproceedings
In: Legal Knowledge and Information Systems, pp. 229–234, IOS Press, 2019.
@inproceedings{vu2019legal,
title = {Legal Text Generation from Abstract Meaning Representation},
author = {Sinh Trong Vu and Minh Le Nguyen and Ken Satoh},
url = {https://ebooks.iospress.nl/pdf/doi/10.3233/FAIA190330},
year = {2019},
date = {2019-01-01},
booktitle = {Legal Knowledge and Information Systems},
pages = {229--234},
publisher = {IOS Press},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
129.
Nguyen, Phuong Minh; Than, Khoat; Nguyen, Minh Le
Marking Mechanism in Sequence-to-sequence Model for Mapping Language to Logical Form Inproceedings
In: 2019 11th International Conference on Knowledge and Systems Engineering (KSE), pp. 1–7, IEEE 2019.
@inproceedings{nguyen2019marking,
title = {Marking Mechanism in Sequence-to-sequence Model for Mapping Language to Logical Form},
author = {Phuong Minh Nguyen and Khoat Than and Minh Le Nguyen},
url = {https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=8919471},
year = {2019},
date = {2019-01-01},
booktitle = {2019 11th International Conference on Knowledge and Systems Engineering (KSE)},
pages = {1--7},
organization = {IEEE},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
128.
Trong, Sinh Vu; Le, Minh Nguyen
An Empirical Evaluation of AMR Parsing for Legal Documents Journal Article
In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11717 LNAI, pp. 131-145, 2019, (cited By 0).
Abstract | Links | BibTeX | Tags:
@article{Vu2019131b,
title = {An Empirical Evaluation of AMR Parsing for Legal Documents},
author = {Sinh Vu Trong and Minh Nguyen Le},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85075556833&doi=10.1007%2f978-3-030-31605-1_11&partnerID=40&md5=43113718dc9930d7c0fa59501b957825},
doi = {10.1007/978-3-030-31605-1_11},
year = {2019},
date = {2019-01-01},
journal = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)},
volume = {11717 LNAI},
pages = {131-145},
abstract = {Many approaches have been proposed to tackle the problem of Abstract Meaning Representation (AMR) parsing, help solving various natural language processing issues recently. In our paper, we provide an overview of different methods in AMR parsing and their performances when analyzing legal documents. We conduct experiments of different AMR parsers on our annotated dataset extracted from the English version of Japanese Civil Code. Our results show the limitations as well as open a room for improvements of current parsing techniques when applying in this non-trivial domain. © 2019, Springer Nature Switzerland AG.},
note = {cited By 0},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Many approaches have been proposed to tackle the problem of Abstract Meaning Representation (AMR) parsing, help solving various natural language processing issues recently. In our paper, we provide an overview of different methods in AMR parsing and their performances when analyzing legal documents. We conduct experiments of different AMR parsers on our annotated dataset extracted from the English version of Japanese Civil Code. Our results show the limitations as well as open a room for improvements of current parsing techniques when applying in this non-trivial domain. © 2019, Springer Nature Switzerland AG.
127.
Chen, Laifu; Nguyen, Minh Le
Sentence selective neural extractive summarization with reinforcement learning Conference
2019, (cited By 5).
Abstract | Links | BibTeX | Tags:
@conference{Chen2019,
title = {Sentence selective neural extractive summarization with reinforcement learning},
author = {Laifu Chen and Minh Le Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85077067618&doi=10.1109%2fKSE.2019.8919490&partnerID=40&md5=b1283c8536e749cd75370549da3f2e3e},
doi = {10.1109/KSE.2019.8919490},
year = {2019},
date = {2019-01-01},
journal = {Proceedings of 2019 11th International Conference on Knowledge and Systems Engineering, KSE 2019},
abstract = {In this work we employed a common Recurrent Neural Network (RNN) based sequence model for single document summarization, composed of encoder-extractor hierarchical network architecture. We develop a sentence level selective encoding mechanism to select important feature before extracting sentences, and use a novel reinforcement learning based training algorithm to extend the sequence model. Besides, for single document extractive summarization task, most of researchers only pay attention to the main part of document. We analyze and explore the side information such as the headline and image caption in both CNN and Daily Mail news datasets. Empirical experiment results show the effect that our model outperforms the baseline model, and can be comparable with the state-of-the-art extractive systems when automatically evaluated in the ROUGE metric. The statistics analysis of the data set verifies our experiment results. © 2019 IEEE.},
note = {cited By 5},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
In this work we employed a common Recurrent Neural Network (RNN) based sequence model for single document summarization, composed of encoder-extractor hierarchical network architecture. We develop a sentence level selective encoding mechanism to select important feature before extracting sentences, and use a novel reinforcement learning based training algorithm to extend the sequence model. Besides, for single document extractive summarization task, most of researchers only pay attention to the main part of document. We analyze and explore the side information such as the headline and image caption in both CNN and Daily Mail news datasets. Empirical experiment results show the effect that our model outperforms the baseline model, and can be comparable with the state-of-the-art extractive systems when automatically evaluated in the ROUGE metric. The statistics analysis of the data set verifies our experiment results. © 2019 IEEE.
126.
Nguyen, Ha-Thanh; Nguyen, Le-Minh
Swarm filter - A simple deep learning component inspired by swarm concept Conference
vol. 2019-November, 2019, (cited By 0).
Abstract | Links | BibTeX | Tags:
@conference{Nguyen20191541,
title = {Swarm filter - A simple deep learning component inspired by swarm concept},
author = {Ha-Thanh Nguyen and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85081086391&doi=10.1109%2fICTAI.2019.00221&partnerID=40&md5=c6de3d885820f4d0a9027c50823713a1},
doi = {10.1109/ICTAI.2019.00221},
year = {2019},
date = {2019-01-01},
journal = {Proceedings - International Conference on Tools with Artificial Intelligence, ICTAI},
volume = {2019-November},
pages = {1541-1545},
abstract = {Swarm is a research topic not only of biologists but also for computer scientists for years. With the idea of swarm intelligence in nature, optimal algorithms are proposed to solve different problems. In addition to the proactive aspect, a swarm can provide useful hints for identification problems. There are features that only exist when an individual belongs to a swarm. An idea came to us, deep learning networks have the ability to automatically select features, so they can extract the characteristics of a swarm for identification problems. This is a new idea in the combination of swarm characteristic with deep learning model. The previous studies combined swarm intelligence with neural networks to find the optimal parameters and architecture for the model. When performing our experiments, we were surprised that this simple architecture got a state-of-the-art result. This interesting discovery can be applied to other tasks using deep learning. © 2019 IEEE.},
note = {cited By 0},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Swarm is a research topic not only of biologists but also for computer scientists for years. With the idea of swarm intelligence in nature, optimal algorithms are proposed to solve different problems. In addition to the proactive aspect, a swarm can provide useful hints for identification problems. There are features that only exist when an individual belongs to a swarm. An idea came to us, deep learning networks have the ability to automatically select features, so they can extract the characteristics of a swarm for identification problems. This is a new idea in the combination of swarm characteristic with deep learning model. The previous studies combined swarm intelligence with neural networks to find the optimal parameters and architecture for the model. When performing our experiments, we were surprised that this simple architecture got a state-of-the-art result. This interesting discovery can be applied to other tasks using deep learning. © 2019 IEEE.
125.
Trang, Luu; Minh, Nguyen Le; Thuy, Nguyen Thanh
Message from the KSE’19 General & TPC Chairs Journal Article
In: 2019.
BibTeX | Tags:
@article{trangmessage,
title = {Message from the KSE’19 General & TPC Chairs},
author = {Luu Trang and Nguyen Le Minh and Nguyen Thanh Thuy},
year = {2019},
date = {2019-01-01},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
124.
Tran, Vu D; Nguyen, Minh L; Shirai, Kiyoaki; Satoh, Ken
2019, (cited By 0).
Abstract | Links | BibTeX | Tags:
@conference{Tran2019,
title = {An approach of rhetorical status recognition for judgments in court documents using deep learning models},
author = {Vu D Tran and Minh L Nguyen and Kiyoaki Shirai and Ken Satoh},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85077023547&doi=10.1109%2fKSE.2019.8919370&partnerID=40&md5=178cbadf39276c8ec3a4b3df9019f41f},
doi = {10.1109/KSE.2019.8919370},
year = {2019},
date = {2019-01-01},
journal = {Proceedings of 2019 11th International Conference on Knowledge and Systems Engineering, KSE 2019},
abstract = {In a court document, the rhetorical status of a sentence conveys the intention of the sentence, whether is is a claim or contains supporting evidences, thus, is beneficial to court document processing systems, for example, court document retrieval systems. Besides, rhetorical structure analysis has high-impact applications in natural language processing, for instances, text summarization, sentiment analysis, question answering. The output structures of the analysis contain high-level relationship between clauses and so provides valuable information. Despite of a wide range of applications and the necessity for automatic court document processing, automatic rhetorical structure analysis has not been well noticed in the legal domain. We propose to use deep learning models for tackling the task of recognizing the rhetorical status of each sentence in a court document. Deep learning has been shown effective towards natural language processing tasks including discourse analysis. We have achieved promising results for the task, which suggests the applicability of artificial neural module embedding rhetorical information for other tasks, for example, summarization and information retrieval. © 2019 IEEE.},
note = {cited By 0},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
In a court document, the rhetorical status of a sentence conveys the intention of the sentence, whether is is a claim or contains supporting evidences, thus, is beneficial to court document processing systems, for example, court document retrieval systems. Besides, rhetorical structure analysis has high-impact applications in natural language processing, for instances, text summarization, sentiment analysis, question answering. The output structures of the analysis contain high-level relationship between clauses and so provides valuable information. Despite of a wide range of applications and the necessity for automatic court document processing, automatic rhetorical structure analysis has not been well noticed in the legal domain. We propose to use deep learning models for tackling the task of recognizing the rhetorical status of each sentence in a court document. Deep learning has been shown effective towards natural language processing tasks including discourse analysis. We have achieved promising results for the task, which suggests the applicability of artificial neural module embedding rhetorical information for other tasks, for example, summarization and information retrieval. © 2019 IEEE.
123.
Tran, Nhu-Thuat; Nguyen, Le-Minh; Phan, Xuan-Hieu; others,
Learning to transform Vietnamese natural language queries into SQL commands Conference
2019, (cited By 2).
Abstract | Links | BibTeX | Tags:
@conference{Vuong2019,
title = {Learning to transform Vietnamese natural language queries into SQL commands},
author = {Nhu-Thuat Tran and Le-Minh Nguyen and Xuan-Hieu Phan and others},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85077042704&doi=10.1109%2fKSE.2019.8919393&partnerID=40&md5=d22d5f85cda03484489e0550da616e08},
doi = {10.1109/KSE.2019.8919393},
year = {2019},
date = {2019-01-01},
journal = {Proceedings of 2019 11th International Conference on Knowledge and Systems Engineering, KSE 2019},
abstract = {In the field of data management, users traditionally manipulates their data using structured query language (SQL). However, this method requires an understanding of relational database, data schema, and SQL syntax as well as the way it works. Database manipulation using natural language, therefore, is much more convenient since any normal user can interact with their data without a background of database and SQL. This is, however, really tough because transforming natural language commands into SQL queries is a challenging task in natural language processing and understanding. In this paper, we propose a novel two-phase approach to automatically analyzing and converting natural language queries into the corresponding SQL forms. In our approach, the first phase is component segmentation which identifies primary clauses in SQL such as SELECT, FROM, WHERE, ORDER BY, etc. The second phase is slot- filling that helps extract sub-components for each primary clause such as SELECT column(s), SELECT aggregation operation, etc. We carefully conducted an empirical evaluation for our method using conditional random fields (CRFs) on a medium-sized corpus of natural language queries in Vietnamese, and have achieved promising results with an average accuracy of more than 90%. © 2019 IEEE.},
note = {cited By 2},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
In the field of data management, users traditionally manipulates their data using structured query language (SQL). However, this method requires an understanding of relational database, data schema, and SQL syntax as well as the way it works. Database manipulation using natural language, therefore, is much more convenient since any normal user can interact with their data without a background of database and SQL. This is, however, really tough because transforming natural language commands into SQL queries is a challenging task in natural language processing and understanding. In this paper, we propose a novel two-phase approach to automatically analyzing and converting natural language queries into the corresponding SQL forms. In our approach, the first phase is component segmentation which identifies primary clauses in SQL such as SELECT, FROM, WHERE, ORDER BY, etc. The second phase is slot- filling that helps extract sub-components for each primary clause such as SELECT column(s), SELECT aggregation operation, etc. We carefully conducted an empirical evaluation for our method using conditional random fields (CRFs) on a medium-sized corpus of natural language queries in Vietnamese, and have achieved promising results with an average accuracy of more than 90%. © 2019 IEEE.
2018
122.
Nguyen, Huy Tien; Nguyen, Minh Le
Multilingual opinion mining on YouTube--A convolutional N-gram BiLSTM word embedding Journal Article
In: Information Processing & Management, vol. 54, no. 3, pp. 451–462, 2018.
BibTeX | Tags:
@article{nguyen2018multilingual,
title = {Multilingual opinion mining on YouTube--A convolutional N-gram BiLSTM word embedding},
author = {Huy Tien Nguyen and Minh Le Nguyen},
year = {2018},
date = {2018-01-01},
journal = {Information Processing & Management},
volume = {54},
number = {3},
pages = {451--462},
publisher = {Pergamon},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
121.
Do, Khac-Phong; Nguyen, Le-Minh
A study on integrating distinct classifiers with bidirectional LSTM for Slot Filling task Conference
2018, (cited By 0).
Abstract | Links | BibTeX | Tags:
@conference{Do2018312,
title = {A study on integrating distinct classifiers with bidirectional LSTM for Slot Filling task},
author = {Khac-Phong Do and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85060380806&doi=10.1109%2fKSE.2018.8573382&partnerID=40&md5=4ffb4c3c4a121e21f0578f94427ea9ef},
doi = {10.1109/KSE.2018.8573382},
year = {2018},
date = {2018-01-01},
journal = {Proceedings of 2018 10th International Conference on Knowledge and Systems Engineering, KSE 2018},
pages = {312-317},
abstract = {In spite of being investigated for decades, in Spoken Language Understanding, slot filling task perceived as sequential labeling in a specific domain is still challenging and attractive to many researchers. For this task, Recurrent Conditional Random Field (RCRF) is a popular model in order to learn latent representations of data which are then utilized as input to a classifier CRF. Our proposed model, in contrast, employed a variant of RNNs, called Long Short-Tem Memory Networks (LSTMs) which, more or less, tackle the downside of RNNs: vanishing gradients. Additionally, we also conducted experiments on the integration of bidirectional LSTM with distinct classifiers, e.g CRFs, SVMs; which then are trained simultaneously. The experimental results show that these combinations are beneficial on both dataset Airline Travel Information System (ATIS) and DARPA Communicator, compared with the state-of-the-art model. © 2018 IEEE.},
note = {cited By 0},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
In spite of being investigated for decades, in Spoken Language Understanding, slot filling task perceived as sequential labeling in a specific domain is still challenging and attractive to many researchers. For this task, Recurrent Conditional Random Field (RCRF) is a popular model in order to learn latent representations of data which are then utilized as input to a classifier CRF. Our proposed model, in contrast, employed a variant of RNNs, called Long Short-Tem Memory Networks (LSTMs) which, more or less, tackle the downside of RNNs: vanishing gradients. Additionally, we also conducted experiments on the integration of bidirectional LSTM with distinct classifiers, e.g CRFs, SVMs; which then are trained simultaneously. The experimental results show that these combinations are beneficial on both dataset Airline Travel Information System (ATIS) and DARPA Communicator, compared with the state-of-the-art model. © 2018 IEEE.
120.
Long, Dang Hoang; Nguyen, Minh-Tien; Bach, Ngo Xuan; Nguyen, Le-Minh; Phuong, Tu Minh
An entailment-based scoring method for content selection in document summarization Conference
2018, (cited By 1).
Abstract | Links | BibTeX | Tags:
@conference{Long2018122,
title = {An entailment-based scoring method for content selection in document summarization},
author = {Dang Hoang Long and Minh-Tien Nguyen and Ngo Xuan Bach and Le-Minh Nguyen and Tu Minh Phuong},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85059963311&doi=10.1145%2f3287921.3287976&partnerID=40&md5=e44073cf17dac5225fe08ebf62f60139},
doi = {10.1145/3287921.3287976},
year = {2018},
date = {2018-01-01},
journal = {ÄCM International Conference Proceeding Series"},
pages = {122-129},
abstract = {This paper introduces a scoring method to improve the quality of content selection in an extractive summarization system. Different from previous models mainly using local information inside sentences such as sentence position or sentence length, our method judges the importance of a sentence based on its own information and the relation between sentences. For the relation between sentences, we utilize textual entailment, a relationship indicating that the meaning of a sentence can be inferred from another one. Unlike previous work on using textual entailment for summarization, we go a step further by looking at aligned words in an entailment sentence pair. Assuming that important words in a salient sentence can be aligned by several words in other sentences, word alignment scores are exploited to compute the entailment score of a sentence. To take advantage of local and neighbor information for facilitating the salient estimation of sentences, we combine entailment scores with sentence position scores. We validate the proposed scoring method with greedy or integer linear programming approaches for extracting summaries. Experiments on three datasets (including DUC 2001 and 2002) in two different domains show that our model obtains competitive ROUGE-scores with state-of-the-art methods for single-document summarization. © 2018 Association for Computing Machinery.},
note = {cited By 1},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
This paper introduces a scoring method to improve the quality of content selection in an extractive summarization system. Different from previous models mainly using local information inside sentences such as sentence position or sentence length, our method judges the importance of a sentence based on its own information and the relation between sentences. For the relation between sentences, we utilize textual entailment, a relationship indicating that the meaning of a sentence can be inferred from another one. Unlike previous work on using textual entailment for summarization, we go a step further by looking at aligned words in an entailment sentence pair. Assuming that important words in a salient sentence can be aligned by several words in other sentences, word alignment scores are exploited to compute the entailment score of a sentence. To take advantage of local and neighbor information for facilitating the salient estimation of sentences, we combine entailment scores with sentence position scores. We validate the proposed scoring method with greedy or integer linear programming approaches for extracting summaries. Experiments on three datasets (including DUC 2001 and 2002) in two different domains show that our model obtains competitive ROUGE-scores with state-of-the-art methods for single-document summarization. © 2018 Association for Computing Machinery.
119.
Nguyen, Khanh Duy Tung; Tuan, Tran Minh; Le, Son Hai; Viet, Anh Phan; Ogawa, Mizuhito; Minh, Nguyen Le
Comparison of Three Deep Learning-based Approaches for IoT Malware Detection Conference
2018, (cited By 12).
Abstract | Links | BibTeX | Tags:
@conference{Nguyen2018382,
title = {Comparison of Three Deep Learning-based Approaches for IoT Malware Detection},
author = {Khanh Duy Tung Nguyen and Tran Minh Tuan and Son Hai Le and Anh Phan Viet and Mizuhito Ogawa and Nguyen Le Minh},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85060368530&doi=10.1109%2fKSE.2018.8573374&partnerID=40&md5=35105f35944599f281f9e4ae56235213},
doi = {10.1109/KSE.2018.8573374},
year = {2018},
date = {2018-01-01},
journal = {Proceedings of 2018 10th International Conference on Knowledge and Systems Engineering, KSE 2018},
pages = {382-387},
abstract = {The development of IoT brings many opportunities but also many challenges. Recently, increasingly more malware has appeared to target IoT devices. Machine learning is one of the typical techniques used in the detection of malware. In this paper, we survey three approaches for IoT malware detection based on the application of convolutional neural networks on different data representations including sequences, images, and assembly code. The comparison was conducted on the task of distinguishing malware from nonmalware. We also analyze the results to assess the pros/cons of each method. © 2018 IEEE.},
note = {cited By 12},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
The development of IoT brings many opportunities but also many challenges. Recently, increasingly more malware has appeared to target IoT devices. Machine learning is one of the typical techniques used in the detection of malware. In this paper, we survey three approaches for IoT malware detection based on the application of convolutional neural networks on different data representations including sequences, images, and assembly code. The comparison was conducted on the task of distinguishing malware from nonmalware. We also analyze the results to assess the pros/cons of each method. © 2018 IEEE.
118.
Le, Tung; Minh, Nguyen Le
Combined Objective Function in Deep Learning Model for Abstractive Summarization Inproceedings
In: Proceedings of the Ninth International Symposium on Information and Communication Technology, pp. 84–91, 2018.
BibTeX | Tags:
@inproceedings{le2018combined,
title = {Combined Objective Function in Deep Learning Model for Abstractive Summarization},
author = {Tung Le and Nguyen Le Minh},
year = {2018},
date = {2018-01-01},
booktitle = {Proceedings of the Ninth International Symposium on Information and Communication Technology},
pages = {84--91},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
117.
Tran, Van-Khanh; Nguyen, Le-Minh
Dual latent variable model for low-resource natural language generation in dialogue systems Conference
2018, (cited By 5).
Abstract | Links | BibTeX | Tags:
@conference{Tran201821,
title = {Dual latent variable model for low-resource natural language generation in dialogue systems},
author = {Van-Khanh Tran and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85072919986&doi=10.18653%2fv1%2fk18-1003&partnerID=40&md5=97328973bf26a7188c94b42eb472548a},
doi = {10.18653/v1/k18-1003},
year = {2018},
date = {2018-01-01},
journal = {CoNLL 2018 - 22nd Conference on Computational Natural Language Learning, Proceedings},
pages = {21-30},
abstract = {Recent deep learning models have shown improving results to natural language generation (NLG) irrespective of providing sufficient annotated data. However, a modest training data may harm such models’ performance. Thus, how to build a generator that can utilize as much of knowledge from a low-resource setting data is a crucial issue in NLG. This paper presents a variational neural-based generation model to tackle the NLG problem of having limited labeled dataset, in which we integrate a variational inference into an encoder-decoder generator and introduce a novel auxiliary auto-encoding with an effective training procedure. Experiments showed that the proposed methods not only outperform the previous models when having sufficient training dataset but also show strong ability to work acceptably well when the training data is scarce. © 2018 Association for Computational Linguistics.},
note = {cited By 5},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Recent deep learning models have shown improving results to natural language generation (NLG) irrespective of providing sufficient annotated data. However, a modest training data may harm such models’ performance. Thus, how to build a generator that can utilize as much of knowledge from a low-resource setting data is a crucial issue in NLG. This paper presents a variational neural-based generation model to tackle the NLG problem of having limited labeled dataset, in which we integrate a variational inference into an encoder-decoder generator and introduce a novel auxiliary auto-encoding with an effective training procedure. Experiments showed that the proposed methods not only outperform the previous models when having sufficient training dataset but also show strong ability to work acceptably well when the training data is scarce. © 2018 Association for Computational Linguistics.
116.
Nguyen12, Minh-Tien; Tran, Duc-Vu; Phan, Viet-Anh; Nguyen, Le-Minh
Towards Social Context Summarization with Convolutional Neural Networks Journal Article
In: 2018.
BibTeX | Tags:
@article{nguyen12towards,
title = {Towards Social Context Summarization with Convolutional Neural Networks},
author = {Minh-Tien Nguyen12 and Duc-Vu Tran and Viet-Anh Phan and Le-Minh Nguyen},
year = {2018},
date = {2018-01-01},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
115.
Tran, Vu; Nguyen, Minh Le; Satoh, Ken
Automatic catchphrase extraction from legal case documents via scoring using deep neural networks Journal Article
In: ärXiv preprint arXiv:1809.05219", 2018.
BibTeX | Tags:
@article{tran2018automatic,
title = {Automatic catchphrase extraction from legal case documents via scoring using deep neural networks},
author = {Vu Tran and Minh Le Nguyen and Ken Satoh},
year = {2018},
date = {2018-01-01},
journal = {ärXiv preprint arXiv:1809.05219"},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
114.
Nguyen, Huy-Tien; Vo, Quan-Hoang; Nguyen, Minh-Le
A Deep Learning Study of Aspect Similarity Recognition Conference
2018, (cited By 2).
Abstract | Links | BibTeX | Tags:
@conference{Nguyen2018181,
title = {A Deep Learning Study of Aspect Similarity Recognition},
author = {Huy-Tien Nguyen and Quan-Hoang Vo and Minh-Le Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85060388874&doi=10.1109%2fKSE.2018.8573326&partnerID=40&md5=730f4e067bbe31227aa03db04a9d01ec},
doi = {10.1109/KSE.2018.8573326},
year = {2018},
date = {2018-01-01},
journal = {Proceedings of 2018 10th International Conference on Knowledge and Systems Engineering, KSE 2018},
pages = {181-186},
abstract = {Äspect Similarity Recognition (ASR) is to identify whether two sentences express one or some aspects in common. This task is useful in review summarization where a summarized review needs to cover all aspects as well as avoid redundancy. To facilitate the application of supervised learning models for this task, we construct a dataset ASRCorpus containing two domains (i.e., LAPTOP and RESTAURANT). The four models i.e., Word Average, CNN, LSTM and BiLSTM) are employed to evaluate the performances of machine learning methods. According to the experimental results, the recurrent neural networks are the most efficient methods. We also analysis some typical samples to identify the task's challenges. © 2018 IEEE."},
note = {cited By 2},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Äspect Similarity Recognition (ASR) is to identify whether two sentences express one or some aspects in common. This task is useful in review summarization where a summarized review needs to cover all aspects as well as avoid redundancy. To facilitate the application of supervised learning models for this task, we construct a dataset ASRCorpus containing two domains (i.e., LAPTOP and RESTAURANT). The four models i.e., Word Average, CNN, LSTM and BiLSTM) are employed to evaluate the performances of machine learning methods. According to the experimental results, the recurrent neural networks are the most efficient methods. We also analysis some typical samples to identify the task's challenges. © 2018 IEEE."
113.
Nguyen, Minh Le
A Study on Social Context Summarization Journal Article
In: 2018.
BibTeX | Tags:
@article{nguyen2018study,
title = {A Study on Social Context Summarization},
author = {Minh Le Nguyen},
year = {2018},
date = {2018-01-01},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
112.
Tran, Van-Khanh; Nguyen, Le-Minh
Adversarial domain adaptation for variational neural language generation in dialogue systems Conference
2018, (cited By 9).
Abstract | Links | BibTeX | Tags:
@conference{Tran20181205,
title = {Adversarial domain adaptation for variational neural language generation in dialogue systems},
author = {Van-Khanh Tran and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85084071344&partnerID=40&md5=bee0f7ea5bb517077eed2fca51a2dc05},
year = {2018},
date = {2018-01-01},
journal = {COLING 2018 - 27th International Conference on Computational Linguistics, Proceedings},
pages = {1205-1217},
abstract = {Domain Adaptation arises when we aim at learning from source domain a model that can perform acceptably well on a different target domain. It is especially crucial for Natural Language Generation (NLG) in Spoken Dialogue Systems when there are sufficient annotated data in the source domain, but there is a limited labeled data in the target domain. How to effectively utilize as much of existing abilities from source domains is a crucial issue in domain adaptation. In this paper, we propose an adversarial training procedure to train a Variational encoder-decoder based language generator via multiple adaptation steps. In this procedure, a model is first trained on a source domain data and then fine-tuned on a small set of target domain utterances under the guidance of two proposed critics. Experimental results show that the proposed method can effectively leverage the existing knowledge in the source domain to adapt to another related domain by using only a small amount of in-domain data. © 2018 COLING 2018 - 27th International Conference on Computational Linguistics, Proceedings. All rights reserved.},
note = {cited By 9},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Domain Adaptation arises when we aim at learning from source domain a model that can perform acceptably well on a different target domain. It is especially crucial for Natural Language Generation (NLG) in Spoken Dialogue Systems when there are sufficient annotated data in the source domain, but there is a limited labeled data in the target domain. How to effectively utilize as much of existing abilities from source domains is a crucial issue in domain adaptation. In this paper, we propose an adversarial training procedure to train a Variational encoder-decoder based language generator via multiple adaptation steps. In this procedure, a model is first trained on a source domain data and then fine-tuned on a small set of target domain utterances under the guidance of two proposed critics. Experimental results show that the proposed method can effectively leverage the existing knowledge in the source domain to adapt to another related domain by using only a small amount of in-domain data. © 2018 COLING 2018 - 27th International Conference on Computational Linguistics, Proceedings. All rights reserved.
111.
Tran, Hong Viet; Nguyen, Van Vinh; Vu, Thuong Huyen; Nguyen, Le Minh
Dependency-based Pre-ordering For English-Vietnamese Statistical Machine Translation Journal Article
In: 2018.
BibTeX | Tags:
@article{tran2018dependency,
title = {Dependency-based Pre-ordering For English-Vietnamese Statistical Machine Translation},
author = {Hong Viet Tran and Van Vinh Nguyen and Thuong Huyen Vu and Le Minh Nguyen},
year = {2018},
date = {2018-01-01},
publisher = {H.: DJHQGHN},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
110.
Nguyen, Minh-Tien; Tran, Duc-Vu; Nguyen, Le-Minh; Phan, Xuan-Hieu
Exploiting user posts for web document summarization Journal Article
In: ÄCM Transactions on Knowledge Discovery from Data", vol. 12, no. 4, 2018, (cited By 4).
Abstract | Links | BibTeX | Tags:
@article{Nguyen2018,
title = {Exploiting user posts for web document summarization},
author = {Minh-Tien Nguyen and Duc-Vu Tran and Le-Minh Nguyen and Xuan-Hieu Phan},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85052604688&doi=10.1145%2f3186566&partnerID=40&md5=fe0600de0f08d9d4493a239a4e43b464},
doi = {10.1145/3186566},
year = {2018},
date = {2018-01-01},
journal = {ÄCM Transactions on Knowledge Discovery from Data"},
volume = {12},
number = {4},
abstract = {Relevant user posts such as comments or tweets of a Web document provide additional valuable information to enrich the content of this document. When creating user posts, readers tend to borrow salient words or phrases in sentences. This can be considered as word variation. This article proposes a framework that models the word variation aspect to enhance the quality of Web document summarization. Technically, the framework consists of two steps: scoring and selection. In the first step, the social information of a Web document such as user posts is exploited to model intra-relations and inter-relations in lexical and semantic levels. These relations are denoted by a mutual reinforcement similarity graph used to score each sentence and user post. After scoring, summaries are extracted by using a ranking approach or concept-based method formulated in the form of Integer Linear Programming. To confirm the efficiency of our framework, sentence and story highlight extraction tasks were taken as a case study on three datasets in two languages, English and Vietnamese. Experimental results show that: (i) the framework can improve ROUGE-scores compared to state-of-the-art baselines of social context summarization and (ii) the combination of the two relations benefits the sentence extraction of single Web documents. © 2018 ACM.},
note = {cited By 4},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Relevant user posts such as comments or tweets of a Web document provide additional valuable information to enrich the content of this document. When creating user posts, readers tend to borrow salient words or phrases in sentences. This can be considered as word variation. This article proposes a framework that models the word variation aspect to enhance the quality of Web document summarization. Technically, the framework consists of two steps: scoring and selection. In the first step, the social information of a Web document such as user posts is exploited to model intra-relations and inter-relations in lexical and semantic levels. These relations are denoted by a mutual reinforcement similarity graph used to score each sentence and user post. After scoring, summaries are extracted by using a ranking approach or concept-based method formulated in the form of Integer Linear Programming. To confirm the efficiency of our framework, sentence and story highlight extraction tasks were taken as a case study on three datasets in two languages, English and Vietnamese. Experimental results show that: (i) the framework can improve ROUGE-scores compared to state-of-the-art baselines of social context summarization and (ii) the combination of the two relations benefits the sentence extraction of single Web documents. © 2018 ACM.
109.
Ngo, Thi-Vinh; Ha, Thanh-Le; Nguyen, Phuong-Thai; Nguyen, Le-Minh
Combining Advanced Methods in Japanese-Vietnamese Neural Machine Translation Conference
2018, (cited By 5).
Abstract | Links | BibTeX | Tags:
@conference{Ngo2018318,
title = {Combining Advanced Methods in Japanese-Vietnamese Neural Machine Translation},
author = {Thi-Vinh Ngo and Thanh-Le Ha and Phuong-Thai Nguyen and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85060402041&doi=10.1109%2fKSE.2018.8573329&partnerID=40&md5=03828bd916b563f2747c16ddb547293b},
doi = {10.1109/KSE.2018.8573329},
year = {2018},
date = {2018-01-01},
journal = {Proceedings of 2018 10th International Conference on Knowledge and Systems Engineering, KSE 2018},
pages = {318-322},
abstract = {Neural machine translation (NMT) systems have recently obtained state-of-the-art in many machine translation systems between popular language pairs because of the availability of data. For low-resourced language pairs, there are few researches in this field due to the lack of bilingual data. In this paper, we attempt to build the first NMT systems for a low-resourced language pair: Japanese-Vietnamese. We have also shown significant improvements when combining advanced methods to reduce the adverse impacts of data sparsity and improve the quality of NMT systems. In addition, we proposed a variant of Byte-Pair Encoding algorithm to perform effective word segmentation for Vietnamese texts and alleviate the rare-word problem that persists in NMT systems. © 2018 IEEE.},
note = {cited By 5},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Neural machine translation (NMT) systems have recently obtained state-of-the-art in many machine translation systems between popular language pairs because of the availability of data. For low-resourced language pairs, there are few researches in this field due to the lack of bilingual data. In this paper, we attempt to build the first NMT systems for a low-resourced language pair: Japanese-Vietnamese. We have also shown significant improvements when combining advanced methods to reduce the adverse impacts of data sparsity and improve the quality of NMT systems. In addition, we proposed a variant of Byte-Pair Encoding algorithm to perform effective word segmentation for Vietnamese texts and alleviate the rare-word problem that persists in NMT systems. © 2018 IEEE.
108.
Nguyen, Truong-Son; Nguyen, Le-Minh; Tojo, Satoshi; Satoh, Ken; Shimazu, Akira
Recurrent neural network-based models for recognizing requisite and effectuation parts in legal texts Journal Article
In: Ärtificial Intelligence and Law", vol. 26, no. 2, pp. 169-199, 2018, (cited By 19).
Abstract | Links | BibTeX | Tags:
@article{Nguyen2018169,
title = {Recurrent neural network-based models for recognizing requisite and effectuation parts in legal texts},
author = {Truong-Son Nguyen and Le-Minh Nguyen and Satoshi Tojo and Ken Satoh and Akira Shimazu},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85044355733&doi=10.1007%2fs10506-018-9225-1&partnerID=40&md5=345553aac7dc5b51e455dcbfa588164c},
doi = {10.1007/s10506-018-9225-1},
year = {2018},
date = {2018-01-01},
journal = {Ärtificial Intelligence and Law"},
volume = {26},
number = {2},
pages = {169-199},
abstract = {This paper proposes several recurrent neural network-based models for recognizing requisite and effectuation (RE) parts in Legal Texts. Firstly, we propose a modification of BiLSTM-CRF model that allows the use of external features to improve the performance of deep learning models in case large annotated corpora are not available. However, this model can only recognize RE parts which are not overlapped. Secondly, we propose two approaches for recognizing overlapping RE parts including the cascading approach which uses the sequence of BiLSTM-CRF models and the unified model approach with the multilayer BiLSTM-CRF model and the multilayer BiLSTM-MLP-CRF model. Experimental results on two Japan law RRE datasets demonstrated advantages of our proposed models. For the Japanese National Pension Law dataset, our approaches obtained an F1 score of 93.27% and exhibited a significant improvement compared to previous approaches. For the Japan Civil Code RRE dataset which is written in English, our approaches produced an F1 score of 78.24% in recognizing RE parts that exhibited a significant improvement over strong baselines. In addition, using external features and in-domain pre-trained word embeddings also improved the performance of RRE systems. © 2018, Springer Science+Business Media B.V., part of Springer Nature.},
note = {cited By 19},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
This paper proposes several recurrent neural network-based models for recognizing requisite and effectuation (RE) parts in Legal Texts. Firstly, we propose a modification of BiLSTM-CRF model that allows the use of external features to improve the performance of deep learning models in case large annotated corpora are not available. However, this model can only recognize RE parts which are not overlapped. Secondly, we propose two approaches for recognizing overlapping RE parts including the cascading approach which uses the sequence of BiLSTM-CRF models and the unified model approach with the multilayer BiLSTM-CRF model and the multilayer BiLSTM-MLP-CRF model. Experimental results on two Japan law RRE datasets demonstrated advantages of our proposed models. For the Japanese National Pension Law dataset, our approaches obtained an F1 score of 93.27% and exhibited a significant improvement compared to previous approaches. For the Japan Civil Code RRE dataset which is written in English, our approaches produced an F1 score of 78.24% in recognizing RE parts that exhibited a significant improvement over strong baselines. In addition, using external features and in-domain pre-trained word embeddings also improved the performance of RRE systems. © 2018, Springer Science+Business Media B.V., part of Springer Nature.
107.
Trieu, Hai-Long; Nguyen, Le-Minh
Enhancing Pivot Translation Using Grammatical and Morphological Information Journal Article
In: Communications in Computer and Information Science, vol. 781, pp. 137-151, 2018, (cited By 0).
Abstract | Links | BibTeX | Tags:
@article{Trieu2018137,
title = {Enhancing Pivot Translation Using Grammatical and Morphological Information},
author = {Hai-Long Trieu and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85044086566&doi=10.1007%2f978-981-10-8438-6_12&partnerID=40&md5=0f7254ea876ab9e477defe33d3890ac8},
doi = {10.1007/978-981-10-8438-6_12},
year = {2018},
date = {2018-01-01},
journal = {Communications in Computer and Information Science},
volume = {781},
pages = {137-151},
abstract = {Pivot translation can be one of the solutions to overcome the problem of unavailable large bilingual corpora for training statistical machine translation models. Nevertheless, the conventional pivot method, which connect source to target phrases via common pivot phrases, lacks some potential connections when pivoting via the surface form of pivot phrases. In this work, we improve the pivot translation method by integrating grammatical and morphological information to connect pivot phrases instead of using only the surface form. Experiments were conducted on several Southeast Asian low-resource language pairs: Indonesian-Vietnamese, Malay-Vietnamese, and Filipino-Vietnamese. By integrating grammatical and morphological information, the proposed method achieved a significant improvement of 0.5 BLEU points. This showed the effectiveness of integrating grammatical and morphological features to pivot translation. © 2018, Springer Nature Singapore Pte Ltd.},
note = {cited By 0},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Pivot translation can be one of the solutions to overcome the problem of unavailable large bilingual corpora for training statistical machine translation models. Nevertheless, the conventional pivot method, which connect source to target phrases via common pivot phrases, lacks some potential connections when pivoting via the surface form of pivot phrases. In this work, we improve the pivot translation method by integrating grammatical and morphological information to connect pivot phrases instead of using only the surface form. Experiments were conducted on several Southeast Asian low-resource language pairs: Indonesian-Vietnamese, Malay-Vietnamese, and Filipino-Vietnamese. By integrating grammatical and morphological information, the proposed method achieved a significant improvement of 0.5 BLEU points. This showed the effectiveness of integrating grammatical and morphological features to pivot translation. © 2018, Springer Nature Singapore Pte Ltd.
106.
Nguyen, Truong-Son; Nguyen, Le-Minh
Nested Named Entity Recognition Using Multilayer Recurrent Neural Networks Journal Article
In: Communications in Computer and Information Science, vol. 781, pp. 233-246, 2018, (cited By 5).
Abstract | Links | BibTeX | Tags:
@article{Nguyen2018233,
title = {Nested Named Entity Recognition Using Multilayer Recurrent Neural Networks},
author = {Truong-Son Nguyen and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85044017100&doi=10.1007%2f978-981-10-8438-6_19&partnerID=40&md5=f80e0cdc713b2f490531a4967e2dc7f7},
doi = {10.1007/978-981-10-8438-6_19},
year = {2018},
date = {2018-01-01},
journal = {Communications in Computer and Information Science},
volume = {781},
pages = {233-246},
abstract = {Many named entities are embedded in others, but current models just only focus on recognizing entities at the top-level. In this paper, we proposed two approaches for the nested named entity recognition task by modeling this task as the multilayer sequence labeling task. Firstly, we propose a model that integrates linguistic features with a neural network to improve the performance of named entity recognition (NER) systems, then we recognize nested named entities by using a sequence of those models in which each model is responsible for predicting named entities at each layer. This approach seems to be inconvenient because we need to train many single models to predict nested named entities. In the second approach, we proposed a novel model, called multilayer recurrent neural networks, to recognize all nested entities at the same time. Experimental results on the Vietnamese data set show that the proposed models outperform previous approaches. Our model yields the state of the art results for Vietnamese with F1 scores of 92.97% at top-level and 74.74% at the nested level. For English, our NER systems also produce better performance. © 2018, Springer Nature Singapore Pte Ltd.},
note = {cited By 5},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Many named entities are embedded in others, but current models just only focus on recognizing entities at the top-level. In this paper, we proposed two approaches for the nested named entity recognition task by modeling this task as the multilayer sequence labeling task. Firstly, we propose a model that integrates linguistic features with a neural network to improve the performance of named entity recognition (NER) systems, then we recognize nested named entities by using a sequence of those models in which each model is responsible for predicting named entities at each layer. This approach seems to be inconvenient because we need to train many single models to predict nested named entities. In the second approach, we proposed a novel model, called multilayer recurrent neural networks, to recognize all nested entities at the same time. Experimental results on the Vietnamese data set show that the proposed models outperform previous approaches. Our model yields the state of the art results for Vietnamese with F1 scores of 92.97% at top-level and 74.74% at the nested level. For English, our NER systems also produce better performance. © 2018, Springer Nature Singapore Pte Ltd.
105.
Nguyen, Minh-Tien; Tran, Duc-Vu; Nguyen, Le-Minh
Social context summarization using user-generated content and third-party sources Journal Article
In: Knowledge-Based Systems, vol. 144, pp. 51-64, 2018, (cited By 8).
Abstract | Links | BibTeX | Tags:
@article{Nguyen201851,
title = {Social context summarization using user-generated content and third-party sources},
author = {Minh-Tien Nguyen and Duc-Vu Tran and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85040514592&doi=10.1016%2fj.knosys.2017.12.023&partnerID=40&md5=5da393634608e4285e058dd9aae24faf},
doi = {10.1016/j.knosys.2017.12.023},
year = {2018},
date = {2018-01-01},
journal = {Knowledge-Based Systems},
volume = {144},
pages = {51-64},
abstract = {In the context of social media, users mutually share their interests of an event mentioned in a Web document. Its content can also be found in different news providers with a writing variation. This paper presents a framework which exploits the support of social context (user-generated content such as comments or tweets and third-party sources such as relevant documents retrieved from a search engine) to extract high-quality summaries. The extraction was formulated in two steps: sentence scoring and selection. The scoring is modeled as a learning to rank problem, which employs Ranking SVM to mutually exploits sentences, user-generated content, and third-party sources in the form of features to cover summary aspects. For the selection, summaries are extracted by using a score-based or voting method. For evaluation, three datasets of sentence and highlight extraction in two languages were taken as a case study. Experimental results indicate that by integrating user-generated content and third-party sources, our framework obtains improvements of ROUGE-scores over state-of-the-art methods for single-document summarization. © 2017},
note = {cited By 8},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
In the context of social media, users mutually share their interests of an event mentioned in a Web document. Its content can also be found in different news providers with a writing variation. This paper presents a framework which exploits the support of social context (user-generated content such as comments or tweets and third-party sources such as relevant documents retrieved from a search engine) to extract high-quality summaries. The extraction was formulated in two steps: sentence scoring and selection. The scoring is modeled as a learning to rank problem, which employs Ranking SVM to mutually exploits sentences, user-generated content, and third-party sources in the form of features to cover summary aspects. For the selection, summaries are extracted by using a score-based or voting method. For evaluation, three datasets of sentence and highlight extraction in two languages were taken as a case study. Experimental results indicate that by integrating user-generated content and third-party sources, our framework obtains improvements of ROUGE-scores over state-of-the-art methods for single-document summarization. © 2017
104.
Nguyen, Huy; Nguyen, Minh-Le
A Deep Neural Architecture for Sentence-Level Sentiment Classification in Twitter Social Networking Journal Article
In: Communications in Computer and Information Science, vol. 781, pp. 15-27, 2018, (cited By 4).
Abstract | Links | BibTeX | Tags:
@article{Nguyen201815,
title = {A Deep Neural Architecture for Sentence-Level Sentiment Classification in Twitter Social Networking},
author = {Huy Nguyen and Minh-Le Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85044048874&doi=10.1007%2f978-981-10-8438-6_2&partnerID=40&md5=566449df15517a1686f0ce4fd86b5d13},
doi = {10.1007/978-981-10-8438-6_2},
year = {2018},
date = {2018-01-01},
journal = {Communications in Computer and Information Science},
volume = {781},
pages = {15-27},
abstract = {This paper introduces a novel deep learning framework including a lexicon-based approach for sentence-level prediction of sentiment label distribution. We propose to first apply semantic rules and then use a Deep Convolutional Neural Network (DeepCNN) for character-level embeddings in order to increase information for word-level embedding. After that, a Bidirectional Long Short-Term Memory network (Bi-LSTM) produces a sentence-wide feature representation from the word-level embedding. We evaluate our approach on three twitter sentiment classification datasets. Experimental results show that our model can improve the classification accuracy of sentence-level sentiment analysis in Twitter social networking. © 2018, Springer Nature Singapore Pte Ltd.},
note = {cited By 4},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
This paper introduces a novel deep learning framework including a lexicon-based approach for sentence-level prediction of sentiment label distribution. We propose to first apply semantic rules and then use a Deep Convolutional Neural Network (DeepCNN) for character-level embeddings in order to increase information for word-level embedding. After that, a Bidirectional Long Short-Term Memory network (Bi-LSTM) produces a sentence-wide feature representation from the word-level embedding. We evaluate our approach on three twitter sentiment classification datasets. Experimental results show that our model can improve the classification accuracy of sentence-level sentiment analysis in Twitter social networking. © 2018, Springer Nature Singapore Pte Ltd.
103.
Tran, Van-Khanh; Nguyen, Le-Minh
Semantic Refinement GRU-Based Neural Language Generation for Spoken Dialogue Systems Journal Article
In: Communications in Computer and Information Science, vol. 781, pp. 63-75, 2018, (cited By 4).
Abstract | Links | BibTeX | Tags:
@article{Tran201863,
title = {Semantic Refinement GRU-Based Neural Language Generation for Spoken Dialogue Systems},
author = {Van-Khanh Tran and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85044096381&doi=10.1007%2f978-981-10-8438-6_6&partnerID=40&md5=269ee643e8c953870350dae493837d4c},
doi = {10.1007/978-981-10-8438-6_6},
year = {2018},
date = {2018-01-01},
journal = {Communications in Computer and Information Science},
volume = {781},
pages = {63-75},
abstract = {Natural language generation (NLG) plays a critical role in spoken dialogue systems. This paper presents a new approach to NLG by using recurrent neural networks (RNN), in which a gating mechanism is applied before RNN computation. This allows the proposed model to generate appropriate sentences. The RNN-based generator can be learned from unaligned data by jointly training sentence planning and surface realization to produce natural language responses. The model was extensively evaluated on four different NLG domains. The results show that the proposed generator achieved better performance on all the NLG domains compared to previous generators. © 2018, Springer Nature Singapore Pte Ltd.},
note = {cited By 4},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Natural language generation (NLG) plays a critical role in spoken dialogue systems. This paper presents a new approach to NLG by using recurrent neural networks (RNN), in which a gating mechanism is applied before RNN computation. This allows the proposed model to generate appropriate sentences. The RNN-based generator can be learned from unaligned data by jointly training sentence planning and surface realization to produce natural language responses. The model was extensively evaluated on four different NLG domains. The results show that the proposed generator achieved better performance on all the NLG domains compared to previous generators. © 2018, Springer Nature Singapore Pte Ltd.
102.
Tran, Viet Hong; Vu, Huyen Thuong; Nguyen, Vinh Van; Nguyen, Minh Le
A classifier-based preordering approach for English-Vietnamese statistical machine translation Journal Article
In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9624 LNCS, pp. 74-87, 2018, (cited By 0).
Abstract | Links | BibTeX | Tags:
@article{Tran201874,
title = {A classifier-based preordering approach for English-Vietnamese statistical machine translation},
author = {Viet Hong Tran and Huyen Thuong Vu and Vinh Van Nguyen and Minh Le Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85044422558&doi=10.1007%2f978-3-319-75487-1_7&partnerID=40&md5=7f4928ceba0dee9ce0a41de525b85699},
doi = {10.1007/978-3-319-75487-1_7},
year = {2018},
date = {2018-01-01},
journal = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)},
volume = {9624 LNCS},
pages = {74-87},
abstract = {Reordering is of essential importance problem for phrase based statistical machine translation (SMT). In this paper, we propose an approach to automatically learn reordering rules as preprocessing step based on a dependency parser in phrase-based statistical machine translation for English to Vietnamese. We used dependency parsing and rules extracting from training the features-rich discriminative classifiers for reordering source-side sentences. We evaluated our approach on English-Vietnamese machine translation tasks, and showed that it outperform the baseline phrase-based SMT system. © Springer International Publishing AG, part of Springer Nature 2018.},
note = {cited By 0},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Reordering is of essential importance problem for phrase based statistical machine translation (SMT). In this paper, we propose an approach to automatically learn reordering rules as preprocessing step based on a dependency parser in phrase-based statistical machine translation for English to Vietnamese. We used dependency parsing and rules extracting from training the features-rich discriminative classifiers for reordering source-side sentences. We evaluated our approach on English-Vietnamese machine translation tasks, and showed that it outperform the baseline phrase-based SMT system. © Springer International Publishing AG, part of Springer Nature 2018.
101.
Tran, Phuoc; Nguyen, Le; Dinh, Dien
PRE-ORDERING OF “DE PHRASE” IN CHINESE-VIETNAMESE MACHINE TRANSLATION Journal Article
In: ICIC express letters. Part B, Applications: an international journal of research and surveys, vol. 9, no. 10, pp. 983–990, 2018.
BibTeX | Tags:
@article{tran2018pre,
title = {PRE-ORDERING OF “DE PHRASE” IN CHINESE-VIETNAMESE MACHINE TRANSLATION},
author = {Phuoc Tran and Le Nguyen and Dien Dinh},
year = {2018},
date = {2018-01-01},
journal = {ICIC express letters. Part B, Applications: an international journal of research and surveys},
volume = {9},
number = {10},
pages = {983--990},
publisher = {ICIC International},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
100.
Tin, Pham Trung; Minh, Nguyen Le
Memory Networks for Fake News Detection Journal Article
In: 2018.
BibTeX | Tags:
@article{tinmemory,
title = {Memory Networks for Fake News Detection},
author = {Pham Trung Tin and Nguyen Le Minh},
year = {2018},
date = {2018-01-01},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
99.
Ittoo, Ashwin; Minh, Nguyen Le; Tojo, Satoshi
Knowledge & Systems Engineering Journal Article
In: Data and Knowledge Engineering, vol. 114, pp. 1–86, 2018.
BibTeX | Tags:
@article{ittoo2018knowledge,
title = {Knowledge & Systems Engineering},
author = {Ashwin Ittoo and Nguyen Le Minh and Satoshi Tojo},
year = {2018},
date = {2018-01-01},
journal = {Data and Knowledge Engineering},
volume = {114},
pages = {1--86},
publisher = {Elsevier},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
98.
Phan, Anh Viet; Chau, Phuong Ngoc; Nguyen, Minh Le; Bui, Lam Thu
Automatically classifying source code using tree-based approaches Journal Article
In: Data and Knowledge Engineering, vol. 114, pp. 12-25, 2018, (cited By 2).
Abstract | Links | BibTeX | Tags:
@article{Phan201812,
title = {Automatically classifying source code using tree-based approaches},
author = {Anh Viet Phan and Phuong Ngoc Chau and Minh Le Nguyen and Lam Thu Bui},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85026469561&doi=10.1016%2fj.datak.2017.07.003&partnerID=40&md5=ca2c752fa936998718b687bd3e957faf},
doi = {10.1016/j.datak.2017.07.003},
year = {2018},
date = {2018-01-01},
journal = {Data and Knowledge Engineering},
volume = {114},
pages = {12-25},
abstract = {Änalyzing source code to solve software engineering problems such as fault prediction, cost, and effort estimation always receives attention of researchers as well as companies. The traditional approaches are based on machine learning, and software metrics obtained by computing standard measures of software projects. However, these methods have faced many challenges due to limitations of using software metrics which were not enough to capture the complexity of programs. To overcome the limitations, this paper aims to solve software engineering problems by exploring information of programs' abstract syntax trees (ASTs) instead of software metrics. We propose two combination models between a tree-based convolutional neural network (TBCNN) and k-Nearest Neighbors (kNN), support vector machines (SVMs) to exploit both structural and semantic ASTs' information. In addition, to deal with high-dimensional data of ASTs, we present several pruning tree techniques which not only reduce the complexity of data but also enhance the performance of classifiers in terms of computational time and accuracy. We survey many machine learning algorithms on different types of program representations including software metrics, sequences, and tree structures. The approaches are evaluated based on classifying 52000 programs written in C language into 104 target labels. The experiments show that the tree-based classifiers dramatically achieve high performance in comparison with those of metrics-based or sequences-based; and two proposed models TBCNN + SVM and TBCNN + kNN rank as the top and the second classifiers. Pruning redundant AST branches leads to not only a substantial reduction in execution time but also an increase in accuracy. © 2017 Elsevier B.V."},
note = {cited By 2},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Änalyzing source code to solve software engineering problems such as fault prediction, cost, and effort estimation always receives attention of researchers as well as companies. The traditional approaches are based on machine learning, and software metrics obtained by computing standard measures of software projects. However, these methods have faced many challenges due to limitations of using software metrics which were not enough to capture the complexity of programs. To overcome the limitations, this paper aims to solve software engineering problems by exploring information of programs' abstract syntax trees (ASTs) instead of software metrics. We propose two combination models between a tree-based convolutional neural network (TBCNN) and k-Nearest Neighbors (kNN), support vector machines (SVMs) to exploit both structural and semantic ASTs' information. In addition, to deal with high-dimensional data of ASTs, we present several pruning tree techniques which not only reduce the complexity of data but also enhance the performance of classifiers in terms of computational time and accuracy. We survey many machine learning algorithms on different types of program representations including software metrics, sequences, and tree structures. The approaches are evaluated based on classifying 52000 programs written in C language into 104 target labels. The experiments show that the tree-based classifiers dramatically achieve high performance in comparison with those of metrics-based or sequences-based; and two proposed models TBCNN + SVM and TBCNN + kNN rank as the top and the second classifiers. Pruning redundant AST branches leads to not only a substantial reduction in execution time but also an increase in accuracy. © 2017 Elsevier B.V."
97.
Phan, Anh Viet; Nguyen, Minh Le; Nguyen, Yen Lam Hoang; Bui, Lam Thu
DGCNN: A convolutional neural network over large-scale labeled graphs Journal Article
In: Neural Networks, vol. 108, pp. 533-543, 2018, (cited By 31).
Abstract | Links | BibTeX | Tags:
@article{Phan2018533,
title = {DGCNN: A convolutional neural network over large-scale labeled graphs},
author = {Anh Viet Phan and Minh Le Nguyen and Yen Lam Hoang Nguyen and Lam Thu Bui},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85055479097&doi=10.1016%2fj.neunet.2018.09.001&partnerID=40&md5=cb429f0a80bb6caf44beacc2b75bdb6d},
doi = {10.1016/j.neunet.2018.09.001},
year = {2018},
date = {2018-01-01},
journal = {Neural Networks},
volume = {108},
pages = {533-543},
abstract = {Exploiting graph-structured data has many real applications in domains including natural language semantics, programming language processing, and malware analysis. A variety of methods has been developed to deal with such data. However, learning graphs of large-scale, varying shapes and sizes is a big challenge for any method. In this paper, we propose a multi-view multi-layer convolutional neural network on labeled directed graphs (DGCNN), in which convolutional filters are designed flexibly to adapt to dynamic structures of local regions inside graphs. The advantages of DGCNN are that we do not need to align vertices between graphs, and that DGCNN can process large-scale dynamic graphs with hundred thousands of nodes. To verify the effectiveness of DGCNN, we conducted experiments on two tasks: malware analysis and software defect prediction. The results show that DGCNN outperforms the baselines, including several deep neural networks. © 2018 Elsevier Ltd},
note = {cited By 31},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Exploiting graph-structured data has many real applications in domains including natural language semantics, programming language processing, and malware analysis. A variety of methods has been developed to deal with such data. However, learning graphs of large-scale, varying shapes and sizes is a big challenge for any method. In this paper, we propose a multi-view multi-layer convolutional neural network on labeled directed graphs (DGCNN), in which convolutional filters are designed flexibly to adapt to dynamic structures of local regions inside graphs. The advantages of DGCNN are that we do not need to align vertices between graphs, and that DGCNN can process large-scale dynamic graphs with hundred thousands of nodes. To verify the effectiveness of DGCNN, we conducted experiments on two tasks: malware analysis and software defect prediction. The results show that DGCNN outperforms the baselines, including several deep neural networks. © 2018 Elsevier Ltd
96.
Yang, Zhen; Chen, Laifu; Nguyen, Minh Le
Regularizing Forward and Backward Decoding to Improve Neural Machine Translation Conference
2018, (cited By 1).
Abstract | Links | BibTeX | Tags:
@conference{Yang201873,
title = {Regularizing Forward and Backward Decoding to Improve Neural Machine Translation},
author = {Zhen Yang and Laifu Chen and Minh Le Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85060372728&doi=10.1109%2fKSE.2018.8573433&partnerID=40&md5=b020bceb0a049e048eab030fd32f74a7},
doi = {10.1109/KSE.2018.8573433},
year = {2018},
date = {2018-01-01},
journal = {Proceedings of 2018 10th International Conference on Knowledge and Systems Engineering, KSE 2018},
pages = {73-78},
abstract = {The common recurrent neural network (RNN) based sequential decoder of the neural machine translation (NMT) model can only translate from one direction, which makes the model overfit in the forward direction and leaves backward information of target sentences unexploited. We propose to use a regularization loss to encourage NMT decoder to exploit bidirectional information of target sentences. Beside of forward decoding, we train an extra set of decoding components to predict translation from backward, and use regularization to enforce the forward and backward hidden states at the same time step have connection. During training phase, the forward hidden states can encode future information from the backward hidden states; while during test phase, we only use the enhanced forward decoding components to translate. Our empirical experiments demonstrated that our approach can significantly improve the performance on WMT German-English and English-Chinese translation tasks in terms of NIST and BLEU score. © 2018 IEEE.},
note = {cited By 1},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
The common recurrent neural network (RNN) based sequential decoder of the neural machine translation (NMT) model can only translate from one direction, which makes the model overfit in the forward direction and leaves backward information of target sentences unexploited. We propose to use a regularization loss to encourage NMT decoder to exploit bidirectional information of target sentences. Beside of forward decoding, we train an extra set of decoding components to predict translation from backward, and use regularization to enforce the forward and backward hidden states at the same time step have connection. During training phase, the forward hidden states can encode future information from the backward hidden states; while during test phase, we only use the enhanced forward decoding components to translate. Our empirical experiments demonstrated that our approach can significantly improve the performance on WMT German-English and English-Chinese translation tasks in terms of NIST and BLEU score. © 2018 IEEE.
95.
Doan, Xuan-Dung; Dang, Trung-Thanh; Nguyen, Minh Le
Effectiveness of character language model for vietnamese named entity recognition Conference
2018, (cited By 0).
Abstract | Links | BibTeX | Tags:
@conference{Doan2018157,
title = {Effectiveness of character language model for vietnamese named entity recognition},
author = {Xuan-Dung Doan and Trung-Thanh Dang and Minh Le Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85090187287&partnerID=40&md5=d7832d92e27be876c8cd2efd2c7cb498},
year = {2018},
date = {2018-01-01},
journal = {Proceedings of the 32nd Pacific Asia Conference on Language, Information and Computation, PACLIC 2018},
pages = {157-163},
abstract = {Recently, many studies indicate that character language model can capture syntacticsemantic word features, resulting in state-ofthe-art performance in typical NLP sequencelabeling tasks. This paper shows the effectiveness of character language model for Vietnamese Named Entity Recognition by comparing several methods. We evaluate the proposed model on the VLSP 2016 dataset andour own VTNER dataset. Experimental resultsshow that our model is the current state-of-theart end-to-end obtains for the task. © 2018 by the authors.},
note = {cited By 0},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Recently, many studies indicate that character language model can capture syntacticsemantic word features, resulting in state-ofthe-art performance in typical NLP sequencelabeling tasks. This paper shows the effectiveness of character language model for Vietnamese Named Entity Recognition by comparing several methods. We evaluate the proposed model on the VLSP 2016 dataset andour own VTNER dataset. Experimental resultsshow that our model is the current state-of-theart end-to-end obtains for the task. © 2018 by the authors.
94.
Nguyen, Huy Thanh; Nguyen, Minh Le
Effective attention networks for aspect-level sentiment classification Inproceedings
In: 2018 10th International Conference on Knowledge and Systems Engineering (KSE), pp. 25–30, IEEE 2018.
BibTeX | Tags:
@inproceedings{nguyen2018effective,
title = {Effective attention networks for aspect-level sentiment classification},
author = {Huy Thanh Nguyen and Minh Le Nguyen},
year = {2018},
date = {2018-01-01},
booktitle = {2018 10th International Conference on Knowledge and Systems Engineering (KSE)},
pages = {25--30},
organization = {IEEE},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
2017
93.
Son, Nguyen Truong; Phan, Viet-Anh; Nguyen, Le Minh
Recognizing entailments in legal texts using sentence encoding-based and decomposable attention models. Inproceedings
In: COLIEE@ ICAIL, pp. 31–42, 2017.
BibTeX | Tags:
@inproceedings{son2017recognizing,
title = {Recognizing entailments in legal texts using sentence encoding-based and decomposable attention models.},
author = {Nguyen Truong Son and Viet-Anh Phan and Le Minh Nguyen},
year = {2017},
date = {2017-01-01},
booktitle = {COLIEE@ ICAIL},
pages = {31--42},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
92.
Nguyen-Van, Hao; Nguyen, Minh; Pham-Nguyen, Loan
An adaptive DC-DC converter for loading circuit of li-ion battery charger Inproceedings
In: 2017 7th International Conference on Integrated Circuits, Design, and Verification (ICDV), pp. 100–103, IEEE 2017.
BibTeX | Tags:
@inproceedings{nguyen2017adaptive,
title = {An adaptive DC-DC converter for loading circuit of li-ion battery charger},
author = {Hao Nguyen-Van and Minh Nguyen and Loan Pham-Nguyen},
year = {2017},
date = {2017-01-01},
booktitle = {2017 7th International Conference on Integrated Circuits, Design, and Verification (ICDV)},
pages = {100--103},
organization = {IEEE},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
91.
Tran, Van-Khanh; Nguyen, Van-Tao; Nguyen, Le-Minh
Enhanced semantic refinement gate for RNN-based neural language generator Conference
vol. 2017-January, 2017, (cited By 0).
Abstract | Links | BibTeX | Tags:
@conference{Tran2017172,
title = {Enhanced semantic refinement gate for RNN-based neural language generator},
author = {Van-Khanh Tran and Van-Tao Nguyen and Le-Minh Nguyen},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85043694420&doi=10.1109%2fKSE.2017.8119454&partnerID=40&md5=d3f463e640b3dedd9bb8a232ba1feb26},
doi = {10.1109/KSE.2017.8119454},
year = {2017},
date = {2017-01-01},
journal = {Proceedings - 2017 9th International Conference on Knowledge and Systems Engineering, KSE 2017},
volume = {2017-January},
pages = {172-178},
abstract = {Natural language generation (NLG) plays an important role in a Spoken Dialogue System. Recurrent Neural Network (RNN)-based approaches have shown promising in tackling NLG tasks. This paper presents approaches to enhance gating mechanism applied for RNN-based natural language generator, in which an attentive dialog act representation is introduced, and two gating mechanisms are proposed to semantically gate input sequences before RNN computation. The proposed RNN-based generators can be learned from unaligned data by jointly training both sentence planning and surface realization to produce natural language responses. The model was extensively evaluated on four different NLG domains. The results show that the proposed generators achieved better performance on all the NLG domains in comparison to the previous generators. © 2017 IEEE.},
note = {cited By 0},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Natural language generation (NLG) plays an important role in a Spoken Dialogue System. Recurrent Neural Network (RNN)-based approaches have shown promising in tackling NLG tasks. This paper presents approaches to enhance gating mechanism applied for RNN-based natural language generator, in which an attentive dialog act representation is introduced, and two gating mechanisms are proposed to semantically gate input sequences before RNN computation. The proposed RNN-based generators can be learned from unaligned data by jointly training both sentence planning and surface realization to produce natural language responses. The model was extensively evaluated on four different NLG domains. The results show that the proposed generators achieved better performance on all the NLG domains in comparison to the previous generators. © 2017 IEEE.