VNU-UET Repository: No conditions. Results ordered -Date Deposited. 2024-03-28T10:22:56ZEPrintshttp://eprints.uet.vnu.edu.vn/images/sitelogo.pnghttps://eprints.uet.vnu.edu.vn/eprints/2021-08-15T08:42:28Z2021-08-15T08:42:28Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4604This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/46042021-08-15T08:42:28ZIterative Multilingual Neural Machine Translation for Less-Common and Zero-Resource Language PairsResearch on providing machine translation systems for unseen language pairs is gaining increasing attention in recent years. However, the quality of their systems is poor for most language pairs, especially for less-common pairs such as Khmer-Vietnamese. In this paper, we show a simple iterative traininggenerating-filtering-training process that utilizes all available pivot parallel data to generate synthetic data for unseen directions. In addition, we propose a filtering method based on word alignments and the longest parallel phrase to filter out noise sentence pairs in the synthetic data. Experiment results on zero-shot Khmer→Vietnamese and Indonesian→Vietnamese directions show that our proposed model outperforms some strong baselines and achieves a promising result under the zero-resource condition on ALT benchmarks. Besides, the results also indicate that our model can easily improve their quality with a small amount of real parallel data.Minh Thuan Nguyenthuannm@vnu.edu.vnPhuong Thai Nguyenthainp@vnu.edu.vnVan Vinh Nguyenvinhnv@vnu.edu.vnMinh Cong Nguyen Hoangminhcongnguyen1508@gmail.com2021-06-28T02:26:30Z2021-06-28T02:26:30Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4517This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/45172021-06-28T02:26:30ZAutomatic Extraction of Analysis Classes from Use CasesMinh Hue ChuDuc Hanh Danghanhdd@vnu.edu.vn2021-06-20T05:06:54Z2021-06-20T05:06:54Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4491This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/44912021-06-20T05:06:54ZOverview of VLSP RelEx shared task: A Data Challenge
for Semantic Relation Extraction from Vietnamese NewsThis paper reports the overview of RelEx shared task for semantic relation extraction from Vietnamese News, which is hosted at the seventh annual workshop on Vietnamese Language and Speech Processing (VLSP 2020). This task focuses on classifying entity pairs in Vietnamese News text into four different, non-overlapping categories of semantic relations defined in advance. In order to generate a fair benchmark, we build a human-annotated dataset of 1,056 documents and 5,900 instances of semantic relations, collected from Vietnamese News in several domains. All models will be evaluated in terms of macro- and micro-averaged F1 scores, two typical evaluation metrics for semantic relation extraction problem.Mai Vu Tranvutm@vnu.edu.vnHoang Quynh Lelhquynh@vnu.edu.vn2021-06-18T11:17:42Z2021-06-18T11:17:42Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4481This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/44812021-06-18T11:17:42ZA framework for assume-guarantee regression verification of evolving softwareThis paper presents a framework for verifying evolving component-based software using assume-guarantee logic. The goal is to improve CDNF-based assumption generation method by having local weakest assumptions that can be used more effectively when verifying component-based software in the context of software evolution. For this purpose, we improve the technique for responding to membership queries when generating candidate assumptions. This technique is then integrated into a proposed backtracking algorithm to generate local weakest assumptions. These assumptions are effectively used in rechecking the evolving software by reducing time required for assumption regeneration within the proposed framework. The proposed framework can be applied to verify software that is continually evolving. An implemented tool and experimental results are presented to demonstrate the effectiveness and usefulness of the framework.Hoang Viet Tranvietth2004@gmail.comNgoc Hung Phamhungpn@vnu.edu.vnViet-Ha NguyenToshiaki Aoki2020-12-31T02:00:05Z2020-12-31T02:00:05Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4353This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/43532020-12-31T02:00:05ZQuy trình kiểm tra trùng lặp trong nhóm văn bảnDinh Hieu Vohieuvd@vnu.edu.vnNgọc Sơn NguyễnMinh Tuấn TrầnVăn Sơn Nguyễn2020-12-18T09:07:29Z2020-12-18T09:07:29Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3983This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/39832020-12-18T09:07:29ZExploiting document graphs for inter-sentence relation extractionBackground: Most previous relation extraction (RE) studies have focused on intra-sentence relations and haveignored relations that span sentences, i.e. inter-sentence relations. Such relations connect entities at thedocument level rather than as relational facts in a single sentence. Extracting facts that are expressed acrosssentences leads to some challenges and requires different approaches than those usually applied in recentintra-sentence relation extraction. Despite recent results, there are still limitations to be overcome.Results:We present a novel representation for a sequence of consecutive sentences, namely documentsub-graph, to extract inter-sentence relations. Experiments on the BioCreative V Chemical-Disease Relationcorpus demonstrate the advantages and robustness of our novel system to extract both intra- andinter-sentence relations in biomedical literature abstracts. The experimental results are comparable tostate-of-the-art approaches and show the potential by demonstrating the effectiveness of graphs, deeplearning-based model and other processing techniques. Experiments were also carried out to verify therationality and impact of various additional information and model components.Conclusions:Our proposed graph-based representation helps to extract∼50% of inter-sentence relations andboosts the model performance on both precision and recall compared to the baseline model.Duy Cat Cancatcd@vnu.edu.vn2020-12-17T08:27:11Z2020-12-17T08:27:11Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4284This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42842020-12-17T08:27:11ZLemma Weakening for State Machine Invariant ProofsLemma conjecture is one of the most challenging tasks in theorem proving. The paper focuses on invariant properties (or invariants) of state machines. Thus, lemmas are also invariants. To prove that a state predicate p is an invariant of a state machine M, in general, we need to find an inductive invariant q of M such that q(s) implies p(s) for all states s of M. q is often in the form p∧p', and p' is often in the form q1 ∧...∧qn. q1, ..., qn are the lemmas of the proof that p is an invariant of M. The paper proposes a technique called Lemma Weakening (LW). LW replaces qi with qi' such that qi(s) implies qi'(s) for all states s of M, which can make the proof reasonably tractable that may become otherwise unreasonably hard. MCS mutual exclusion protocol is used as an example to demonstrate the power of LW.Dinh Duong Tranduongtd@vnu.edu.vnKazuhiro Ogataogata@jaist.ac.jpDuy Dang Buibddang@jaist.ac.jpParth Guptaparthgupta.iitkgp@gmail.com2020-12-16T09:45:53Z2020-12-16T09:45:53Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4279This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42792020-12-16T09:45:53ZMÔ HÌNH HỆ THỐNG QUẢN TRỊ TIN NHẮN DỰA TRÊN MODULE MEI-CHAT CHO CÁC TRANG WEB CÓ SỐ LƯỢNG NGƯỜI DÙNG LỚNDịch vụ chăm sóc khách hàng trực tuyến trên các trang web đang là một thành phần quan trọng để duy trì hoạt động kinh doanh trên môi trường mạng. Các hộp thoại nhắn tin trực tuyến được nhúng vào các trang web cung cấp khả năng liên lạc giữa người truy cập và quản trị viên trang web. Trong bài báo này này, chúng tôi giới thiệu module mei-chat xây dựng bằng các công nghệ Web tiên tiến nhất như NodeJS, AngularJS, SocketIO. Mei-chat gồm ba thành chính là meiChatCustomer, meiChatAdmin và meiChatBackend. MeiChatBackend được xây dựng bằng Nodejs tích hợp SocketIO với cấu trúc cân bằng tải nhằm đảm bảo số lượng lớn người kết nối. MeiChatAdmin cho phép quản trị viên nhắn tin, xem lịch sử tin nhắn của người truy cập trang web. MeiChatCustomer là hộp thoại tin nhắn có thể dễ dàng tích hợp vào các trang web, cho phép người truy cập nhắn tin với quản trị viên.Thi Nhan NguyenXuan Tung Hoangtunghx@vnu.edu.vnNgoc Dung Buidnbui@utc.edu.vn2020-12-16T09:45:31Z2020-12-16T09:45:31Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4277This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42772020-12-16T09:45:31ZPOSE INVARIANT FACE RECOGNITION SYSTEMFace recognition has become popular in security surveillance systems. These systems work well only when the face is in frontal and the system itself has considerable training data. This paper proposes a method of face recognition with various view using Generative Adversarial Networks (GAN). In this method, a pre-trained model of face is used to generate the training data of different pose variations from several frontal faces using GAN. These faces will be used as input of the FaceNet and Support Vector Machine to be trained to perform the task of face recognition. The experiment results demonstrate the accuracy of the proposed method in the case of lacking the training data. This paper focus on the applicability of Generative Adversarial Network in Face Recognition System by applying the GAN-based method in Face Recognition System to increase the number of training samples and detect the faces of different poses that was not possible with the conventional Face Recognition System. The comparison between Face Recognition System using GAN and conventional Face Recognition System was presented.Dong Dong NguyenNgoc Dung Buidnbui@utc.edu.vnXuan Tung Hoangtunghx@vnu.edu.vnNguyen Cac Tran2020-12-14T05:02:45Z2020-12-14T05:03:05Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4260This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42602020-12-14T05:02:45ZPrivacy-Preserving Visual Content Tagging using Graph Transformer NetworksWith the rapid growth of Internet media, content tagging has become an important topic with many multimedia understanding applications, including efficient organisation and search. Nevertheless, existing visual tagging approaches are susceptible to inherent privacy risks in which private information may be exposed unintentionally. The use of anonymisation and privacy-protection methods is desirable, but with the expense of task performance.Therefore, this paper proposes an end-to-end framework (SGTN) using Graph Transformer and Convolutional Networks to significantly improve classification and privacy preservation of visual data. Especially, we employ several mechanisms such as differential privacy based graph construction and noise-induced graph transformation to protect the privacy of knowledge graphs. Our approach unveils new state-of-the-art on MS-COCO dataset in various semisupervised settings. In addition, we showcase a real experiment in the education domain to address the automation of sensitive document tagging. Experimental results show that our approach achieves an excellent balance of model accuracy and privacy preservation on both public and private datasets. Codes are available at https://github.com/ReML-AI/sgtn.Xuan Son Vusonvx@cs.umu.seDuc Trong Letrongld@vnu.edu.vnChristoffer Edlundchristoffer.edlund@sartorius.comLili Jianglili.jiang@cs.umu.seHoang D. NguyenHarry.Nguyen@glasgow.ac.uk2020-12-13T15:45:38Z2020-12-13T15:46:11Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3967This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/39672020-12-13T15:45:38ZMultimodal Review Generation with Privacy and Fairness AwarenessUsers express their opinions towards entities (e.g., restaurants) via online reviews which can bein diverse forms such as text, ratings, and images. Modeling reviews are advantageous for userbehavior understanding which, in turn, supports various user-oriented tasks such as recommen-dation, sentiment analysis, and review generation. In this paper, we propose MG-PriFair, a multi-modal neural-based framework, which generates personalized reviews with privacy and fairnessawareness. Motivated by the fact that reviews might contain personal information and sentimentbias, we propose a novel differentially private (dp)-embedding model for training privacy guar-anteed embeddings and an evaluation approach for sentiment fairness in the food-review domain.Experiments on our novel review dataset show that MG-PriFair is capable of generating plausiblylong reviews while controlling the amount of exploited user data and using the least sentiment-biased word embeddings. To the best of our knowledge, we are the first to bring user privacy andsentiment fairness into the review generation task. The dataset and source codes are available athttps://github.com/ReML-AI/MG-PriFair.Xuan Son Vusonvx@cs.umu.seThanh Son Nguyenant.sonnt@gmail.comDuc Trong Letrongld@vnu.edu.vnLili Jianglili.jiang@cs.umu.se2020-12-12T15:29:21Z2020-12-12T15:29:21Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4251This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42512020-12-12T15:29:21ZOptimization of IoT Service Deployment In Multi-Layered Cloud-Fog EnvironmentRecently, fog computing, which can be done in proximity to data sources, has emerged as a solution to provide low-latency Quality-of-Service (QoS) for IoT services in complement to centralized cloud with unlimited computing resources. Optimized service deployment on both cloud and fog environments is challenging due to their heterogeneity. Prior works mainly focus on mapping service functions and dependencies directly to physical network. In this paper, we propose a multi-layer mapping mechanism that efficiently deploys multiple IoT services to the appropriate virtual networks in physical infrastructure. We design greedy-based algorithms for solving this NP-hard problem with two phases executed sequentially. Experimental results show our proposed solution can reduce upto 80% of the total service cost compared to the state-of-the-art solutions.Van Do Dangdodv@vnu.edu.vnXuan Tung Hoangtunghx@vnu.edu.vnNguyen Kim-Khoakim-khoa.nguyen@etsmtl.ca2020-12-10T11:00:58Z2020-12-10T11:00:58Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4234This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42342020-12-10T11:00:58ZModelling of Lattice Constants of the Garnet Structured Compounds Using Machine LearningGarnet structured compounds have been widely utilized as the host materials of phosphors. Due to their complicated structures with numerous atoms in a unit cell of the garnet structure, it is difficult to predict their crystal structures accurately by the first principles calculation, especially when they include magnetic elements. It was proposed to predict the structural parameters for spinel compounds by the empirical model employing ionic radii
and electronegativities, which reproduced the experimental structural parameters of spinels successfully. In the present work, we propose a comprehensive and reliable model to explain and/or predict the structural parameters of the garnet structured materials using the machine learning. The lattice parameters of 182 garnet compounds reported in the database are compiled in our dataset to train the model. Using the ionic radii and electronegativities of the constituent elements of considered garnet compounds, we constructed the linear regression model fitted to the garnets' lattice constants in the dataset. The predicted values obtained as a result of the training of the model exhibited high correlation with the actual lattice constants, having the correlation coefficient of 0.988. For each composition in our dataset, the relative error between experimental and calculated lattice constants was less than 1.60%.Masatoshi Kubomasa104k@ruri.waseda.jpHai Chau Nguyenchaunh@vnu.edu.vnSergey BrikTomoyuki Yamamototymmt@waseda.jp2020-12-10T03:41:08Z2020-12-10T03:41:08Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4225This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42252020-12-10T03:41:08ZiK-means: an improvement of the iterative k-means partitioning algorithmThu LeSy Vinh Levinhls@vnu.edu.vnDong Do Ducdongdd@vnu.edu.vnThang Buithangbn@vnu.edu.vnThao Nguyenthaontp@ioit.ac.vn2020-12-10T03:41:00Z2020-12-10T03:41:00Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4223This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42232020-12-10T03:41:00ZA protein alignment partitioning method for protein phylogenetic inferenceThu LeSy Vinh Levinhls@vnu.edu.vn2020-12-10T03:40:53Z2020-12-10T03:40:53Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4222This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42222020-12-10T03:40:53ZPHƯƠNG PHÁP VÀ HỆ THỐNG NHẬP LIỆU DỰA TRÊN BÀN PHÍM ẢO VÀ CHUYỂN THÀNH ÂM THANHThanh Ha Leltha@vnu.edu.vnThe Duy Buiduybt@vnu.edu.vnThi Duyen NgoDuyennt@vnu.edu.vnVan Tung NguyenThe Hoang Anh NguyenMinh Hoa Nguyen2020-12-10T03:40:44Z2020-12-10T03:40:44Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4219This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42192020-12-10T03:40:44ZFLAVI: An Amino Acid Substitution Model for FlavivirusesThu NguyenLe Sy Vinhvinhls@vnu.edu.vn2020-12-10T03:40:32Z2020-12-10T03:40:32Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4217This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42172020-12-10T03:40:32ZPolyp Segmentation in Colonoscopy Images Using Ensembles of U-Nets with EfficientNet and Asymmetric Similarity Loss FunctionAutomatic polyp detection and segmentation are highly desirable for colon screening due to polyp miss rate by physicians during colonoscopy, which is about 25%. Diagnosis of polyps in colonoscopy videos is a challenging task due to variations in the size and shape of polyps. In this paper, we adapt U-net and evaluate its performance with different modern convolutional neural networks as its encoder for polyp segmentation. One of the major challenges in training networks for polyp segmentation raises when the data are unbalanced, polyp pixels are often much lower in numbers than non-polyp pixels. A trained network with unbalanced data may make predictions with high precision and low recall, being severely biased toward the non-polyp class which is particularly undesired because false negatives are more important than false positives. We propose an asymmetric similarity loss function to address this problem and achieve a much better tradeoff between precision and recall. Finally, we propose an ensemble method for further performance improvement. We evaluate the performance of well-known polyp datasets CVC-ColonDB and ETIS-Larib PolypDB. The best results are 89.13% dice, 79.77% IOU, 90.15% recall, and 86.28% precision. Our proposed method outperforms the state-of-the-art polyp segmentation methods.Thu Hong LeChi Thanh NguyenQuoc Long Trantqlong@vnu.edu.vn2020-12-10T02:44:46Z2020-12-10T02:44:46Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4218This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42182020-12-10T02:44:46ZKey phrase Generation for Vietnamese Administrative Documents: A Collaborative ApproachThi Thu Trang NguyenThi Hai Yen Vuongyenvth_57@vnu.edu.vnVan Lien Tran14020768@vnu.edu.vnLe Minh NguyenXuan Hieu Phanhieupx@vnu.edu.vn2020-12-10T02:41:48Z2020-12-10T02:41:48Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4216This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42162020-12-10T02:41:48ZLearning Neural Textual Representations for Citation RecommendationWith the rapid growth of the scientific literature, manually selecting appropriate citations for a paper is becoming increasingly challenging and time-consuming. While several approaches for automated citation recommendation have been proposed in the recent years, effective document representations for citation recommendation are still elusive to a large extent. For this reason, in this paper we propose a novel approach to citation recommendation which leverages a deep sequential representation of the documents (Sentence-BERT) cascaded with Siamese and triplet networks in a submodular scoring function. To the best of our knowledge, this is the first approach to combine deep representations and submodular selection for a task of citation recommendation. Experiments have been carried out using a popular benchmark dataset - the ACL Anthology Network corpus - and evaluated against baselines and a state-of-the-art approach using metrics such as the MRR and F1-at-k score. The results show that the proposed approach has been able to outperform all the compared approaches in every measured metric.Thanh Binh KieuInigo Jauregi UnanueBao Son Phamsonpb@vnu.edu.vnXuan Hieu Phanhieupx@vnu.edu.vnMassimo Piccardi2020-12-10T02:37:36Z2020-12-10T02:37:36Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4215This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42152020-12-10T02:37:36ZpQMaker: empirically estimating amino acid substitution models in a parallel environmentAmino acid substitution matrices are central to the model-based methods for reconstructing evolutionary trees from amino acid sequences. QMaker is an efficient method for estimating general time-reversible amino acid substitution matrices from a large biological dataset containing thousands of protein alignments using maximum likelihood principle. It allows researchers to build an amino acid substitution model on their own to best fit their subsequent phylogenetic analyses. In this work, we propose an approach to parallelize computation in QMaker, named pQMaker. Moreover, we provide an open-source message passing interface implementation for pQMaker (https://github.com/canhnd58/IQ-TREE/tree/pqmaker) built upon the latest IQ-TREE package. Experiments on benchmark data sets show that our implementation has significant speed gains compared with the original QMaker.Duc Canh Nguyenduccanh9511@gmail.comCao Cuong DangSy Vinh Levinhls@vnu.edu.vnQuang Minh Buim.bui@anu.edu.auThi Diep Hoangdiepht@vnu.edu.vn2020-12-09T14:49:33Z2020-12-11T01:44:19Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4211This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42112020-12-09T14:49:33ZOptimization of IoT Service Deployment In Multi-Layered Cloud-Fog EnvironmentVan Do Dangdodv@vnu.edu.vnXuan Tung Hoangtunghx@vnu.edu.vnMai TranKim Khoa NguyenKim-Khoa.Nguyen@etsmtl.ca2020-12-09T03:20:02Z2020-12-09T03:20:02Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4204This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/42042020-12-09T03:20:02ZSingle Concatenated Input is Better than Indenpendent Multiple-input for CNNs to Predict Chemical-induced Disease Relation from LiteratureChemical compounds (drugs) and diseases are among top searched keywords on the PubMed database of biomedical literature by biomedical researchers all over the world (according to a study in 2009). Working with PubMed is essential for researchers to get insights into drugs’ side effects (chemical-induced disease relations (CDR), which is essential for drug safety and toxicity. It is, however, a catastrophic burden for them as PubMed is a huge database of unstructured texts, growing steadily very fast (~28 millions scientific articles currently, approximately two deposited per minute). As a result, biomedical text mining has been empirically demonstrated its great implications in biomedical research communities. Biomedical text has its own distinct challenging properties, attracting much attetion from natural language processing communities. A large-scale study recently in 2018 showed that incorporating information into indenpendent multiple-input layers outperforms concatenating them into a single input layer (for biLSTM), producing better performance when compared to state-of-the-art CDR classifying models. This paper demonstrates that for a CNN it is vice-versa, in which concatenation is better for CDR classification. To this end, we develop a CNN based model with multiple input concatenated for CDR classification. Experimental results on the benchmark dataset demonstrate its outperformance over other recent state-of-the-art CDR classification models.Thi Quynh Trang Phamtrangptq@vnu.edu.vnManh Thang BuiThanh Hai Danghai.dang@vnu.edu.vn2020-12-08T09:19:00Z2020-12-08T09:47:50Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3712This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/37122020-12-08T09:19:00ZMột mở rộng tập thô phủ và ứng dụngLý thuyết tập thô và các mở rộng của nó như tính toán hạt dựa trên tập thô, tập thô phủ, tâp thô mờ,... ngày càng được phát triển và áp dụng hiệu quả trong nhiều lĩnh vực. Bài báo tập trung giới thiệu khái niệm bảng quyết định phủ, một ví dụ và phân tích về việc xây dựng bảng quyết định phủ. Bài báo cũng đưa ra một gợi ý về việcc ứng dụng bảng quyết định phủ trong hệ tư vấn.Thanh Huyen Phamphamthanhhuyen@daihochalong.edu.vnThuan Hohothuan@vast.ac.vnQuang Thuy Hathuyhq@vnu.edu.vnTri Thanh Nguyenntthanh@vnu.edu.vnXuan Bach Nguyen2020-12-08T09:18:22Z2020-12-08T09:18:22Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4177This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/41772020-12-08T09:18:22ZPhương pháp và thiết bị phân tích nội dung viđeo để phát hiện sự kiện người ngãThanh Ha Leltha@vnu.edu.vnThị Thủy NguyenQuoc Long Trantqlong@vnu.edu.vnDo Van NguyenChi Thanh NguyenViet Anh Nguyenvietanh@cic.com.vnCong Thanh Nguyennc.thanh1995@gmail.com2020-12-08T09:17:50Z2020-12-08T09:17:50Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4176This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/41762020-12-08T09:17:50ZA Deep Wavelet Sparse Autoencoder Method for Online and Automatic Electrooculographical Artifact RemovalThe Hoang Anh NguyenThe Duy Buiduybt@vnu.edu.vnThanh Ha Leltha@vnu.edu.vn2020-12-08T09:17:24Z2020-12-08T09:17:24Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4180This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/41802020-12-08T09:17:24ZA Targeted Topic Model based Multi-Label
Deep Learning Classification Framework
for Aspect-based Opinion MiningRecently, deep Convolutional Neural Network (CNN) model has achieved remarkable results in Natural Language Processing (NLP) tasks, such as information retrieval, relation classification, semantic parsing, sentence modeling and other traditional NLP tasks, etc. On the other hand, topic modeling method has been proved to be effective by exploiting hidden knowledge in a corpus of documents. Motivated from these successes, we propose a framework that takes advantages of closure domain measure to get enriched knowledge from close domains to the dataset of the current task to improve the CNN model, and apply a Targeted Topic Model to take more detailed exploration on each labeled aspect of an opinion. Experimental results on different scenarios show the effectiveness of the proposed framework for multi-label classification task in comparison to other related models on the same Hotel review dataset.Thi Cham Nguyennthicham@hpmu.edu.vnThi Ngan Phamptngan2012@gmail.comHoang Quynh Lelhquynh@gmail.comTri Thanh Nguyenntthanh@vnu.edu.vnHong Nhung Buinhungbth@hvnh.edu.vnQuang Thuy Hathuyhq@vnu.edu.vn2020-12-08T09:17:04Z2020-12-08T09:17:04Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4175This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/41752020-12-08T09:17:04ZA Word + Character Embedding based Relation Extraction Frame for Domain Ontology of Natural Resources and EnvironmentBuilding domain ontology is a challenging problem, and there are many different approaches for domain ontology construction. However, most of these approaches are still mainly using manual methods [1]. Ontology enrichment is a fairly standard approach in domain ontology construction, in which semi-automated methods and automated methods of ontology learning from a derived ontology. Relation extraction is one of the ways for ontology enrichment. Relation extraction techniques include law-based techniques, machine learning-based techniques with three typical methods: supervised learning, semi-supervised learning, and unsupervised learning. This paper proposes a word + character embedding-based relation extraction frame for the Vietnamese domain ontology of natural resources and environment. The model's effect was demonstrated by experiments in the domain of natural resources and the envi-ronment and achieving promising results.Ngoc Vu Nguyenvunnciren@gmail.comMai Vu Tranvutm@vnu.edu.vnHai Chau Nguyenchaunh@vnu.edu.vnQuang Thuy Hathuyhq@vnu.edu.vn2020-12-08T09:16:35Z2020-12-08T09:16:35Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4174This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/41742020-12-08T09:16:35ZA Definition of Covering Based Decision Table and Its Sample ApplicationsCovering based rough set, an extension of the traditional rough set theory, which uses the cover set of the universe set instead of the partition of the universe, has proven to be both theoretical and attractive in terms of applications. Corresponding to the decision table in traditional rough set theory, the concept of covering decision system has been defined. In this paper, we propose a decision table type based on covers, including the condition lattice of covers, and the decision lattice of covers. Two tasks on covering based decision table are also introduced. We also demonstrate the applications of the covering based decision table in collaborative filtering that corresponds to the classification in the traditional decision table, and in constraint based association rule mining to indicate this covering decision table concept has a potential applicationThanh Huyen Phamphamthanhhuyen@daihochalong.edu.vnThi Cam Van Nguyenvanntc@vnu.edu.vnThi Hong Vuonghongvt_57@vnu.edu.vnThuan Hohothuan@vast.ac.vnQuang Thuy Hathuyhq@vnu.edu.vnTri Thanh Nguyenntthanh@vnu.edu.vn2020-12-08T09:16:08Z2020-12-08T09:16:08Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4173This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/41732020-12-08T09:16:08ZĐo lường công bố khoa học: bối cảnh thế giới và liên hệ với Việt NamĐo lường công bố khoa học của cá nhân nhà khoa học, tổ chức khoa học là một chủ đề hấp dẫn theo các khía cạnh khoa học, kinh tế và quản lý. Tuy nhiên, đo lường công bố khoa học cũng gặp không ít thách thức, vì vậy, việc tìm hiểu một cách toàn diện về đo lường công bố khoa học là một công việc cần thiết. Bài viết này cung cấp một khảo sát bước đầu song đủ khái quát về đo lường công bố khoa học, một số thách thức chính và xu hướng hoạt động này trong thời đại số. Một số trao đổi về hoạt động đo lường công bố khoa học tại Việt Nam cũng được trình bày.Quang Thuy Hathuyhq@vnu.edu.vnDinh Hieu Vohieuvd@vnu.edu.vnXuan Hieu Phanhieupx@vnu.edu.vnTri Thanh Nguyenntthanh@vnu.edu.vnHuu Duc Nguyenducnh@vnu.edu.vn2020-12-07T14:07:31Z2020-12-21T09:33:05Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4155This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/41552020-12-07T14:07:31ZPCA-Based Robust Motion Data RecoveryHuman motion tracking is a prevalent technique in many fields. A common difficulty encountered in motion tracking is the corrupted data is caused by detachment of markers in 3D motion data or occlusion in 2D tracking data. Most methods for missing markers problem may quickly become ineffective when gaps exist in the trajectories of multiple markers for an extended duration. In this paper, we propose the principal component eigenspace based gap filling methods that leverage a training sample set for estimation. The proposed method is especially beneficial in the scenario of motion data with less predictable or repeated movement patterns, and that of even missing entire frames within an interval of a sequence. To highlight algorithm robustness, we perform algorithms on twenty test samples for comparison. The experimental results show that our methods are numerical stable and fast to work.Zhuorong LiHongchuan YuHai Dang KieuTung Long VuongJian Jun Zhang2020-12-07T14:07:13Z2020-12-07T14:07:13Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4153This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/41532020-12-07T14:07:13ZReducing Blocking Artifacts in CNN-Based Image Steganography by Additional Loss FunctionsOur work improves the encoded image quality from HiDDeN framework, an end-to-end image steganography based on deep convolution neural network.
In the encoding phase of HiDDeN framework, to embed a message in a cover image, it is required to split the cover image into smaller image blocks and embed the message bits in each block in parallel. These embedded blocks are then combined to form an encoded image that has the same size as the cover image. This image reconstruction process causes artifacts that appear on the boundaries of the blocks. This can be explained by the fact that when message bits are embedded in the image blocks, the pixel-level information of each image block is unequally alternated.
In order to reduce block artifacts,
in this work we propose a blocking loss as an additional objective function in HiDDeN framework. This loss measures the difference between encoded images and modified versions of the cover images.
The proposed method is evaluated on COCO 2014 and BOSS datasets and the experimental results show the effectiveness in reducing the block artifacts that appeared in the encoded images of HiDDeN framework. This has an important impact on increasing the invisibility or transparency of the steganography system. In addition, the experimental result on secrecy of the proposed method also indicates the same performance as the HiDDeN pipeline.Tuan Dung Phamphamtuandung12@gmail.comViet Cuong Tacuongtv@vnu.edu.vnThi Thanh Thuy PhamThanh Ha Leltha@vnu.edu.vn2020-12-07T04:19:27Z2020-12-07T04:19:27Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4146This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/41462020-12-07T04:19:27ZGenerative Software Module Development for Domain-Driven Design with Annotation-Based Domain Specific LanguageMinh Duc LeDuc Hanh Danghanhdd@vnu.edu.vnViet Ha Nguyenhanv@vnu.edu.vn2020-12-05T13:34:53Z2020-12-21T09:26:41Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4143This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/41432020-12-05T13:34:53ZCompression Artifacts Image Patch database for
Perceptual Quality AssessmentGround truth is one of the most important component for training, testing, and benchmarking algorithms for
objective quality assessment. In this paper, we propose an image
patch quality database with compression artifacts. We create
a new database of image patches with High Efficiency Video
Coding (HEVC) compression artifacts. Then, the subjective test
is conducted in a controlled environment to obtain the ground
truth of image patch quality, where we collect differential mean
opinion scores (DMOS) from a larger amount of observers.
Finally, the rank order correlation factors between DMOS and a
set of popular image quality metrics are calculated and presented.
The proposed database is expected for learning patch based IQA
model for block size in video rate-distortion optimization.Tung Pham ThanhChau Ma ThiTuan Nguyen ManhLinh Le DinhHa Le Thanh2020-11-20T16:11:36Z2020-11-20T16:11:36Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4084This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40842020-11-20T16:11:36ZNhững yếu tố quan trọng trong Chiến lược quốc gia về trí tuệ nhân tạoChuyển đổi số và trí tuệ nhân tạo (TTNT) là các quá trình tất yếu, song cũng rất phức tạp đối với các tổ chức và các quốc gia. Nhóm tác giả đã chỉ ra 3 yếu tố (trưởng thành số, kiến trúc tổ chức chuyển đổi cân đối - CĐCĐ, tài năng con người về TTNT) đóng vai trò quan trọng đặc biệt trong chiến lược quốc gia về TTNT, trong chiến lược và sáng kiến chuyển đổi số và TTNT của tổ chức.Quang Thuy Hathuyhq@vnu.edu.vnBao Son Phamsonpb@vnu.edu.vnXuan Hieu Phanhieupx@vnu.edu.vnTrong Hieu Tranhieutt@vnu.edu.vnMai Vu Tranvutm@vnu.edu.vnTri Thanh Nguyenntthanh@vnu.edu.vnVan Dung Phamdungpv@vnu.edu.vnThe Hiep Tranhieptt@vnu.edu.vn2020-11-11T07:42:48Z2020-11-11T07:42:48Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4083This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40832020-11-11T07:42:48ZPractical approach to access the impact of global variables on program parallelismGlobal variables may have a significant impact on preventing programs from automatic parallelism. This paper introduces a practical approach to measure the effect of global variables on program parallelism. First, we conduct static data dependence analysis among program variables and represent such dependencies by a Variable Dependence Graph. Then, we analyze this graph for measuring and identifying which global variables have a significant impact on program parallelism. To evaluate this approach, we conduct experiments on 20 benchmark programs and an industrial application. The experimental results show that half of the studied programs contain large impact variables which may be the cause of preventing programs from parallel execution.Thu Trang Nguyentrang.nguyen@vnu.edu.vnHue Nguyen17021156@vnu.edu.vnQuang Cuong Buibqcuong@vnu.edu.vnNgoc Hung Phamhungpn@vnu.edu.vnDinh Hieu Vohieuvd@vnu.edu.vnShigeki Takeuchitakeuchi.s@gaio.co.jp2020-10-13T08:18:06Z2020-10-13T08:18:28Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4068This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40682020-10-13T08:18:06ZBisimulation and bisimilarity for fuzzy description logics under the Gödel semanticsDescription logics (DLs) are a suitable formalism for representing knowledge about domains in which objects are described not only by attributes but also by binary relations between objects. Fuzzy extensions of DLs can be used for such domains when data and knowledge about them are vague and imprecise. One of the possible ways to specify classes of objects in such domains is to use concepts in fuzzy DLs. As DLs are variants of modal logics, indiscernibility in DLs is characterized by bisimilarity. The bisimilarity relation of an interpretation is the largest auto-bisimulation of that interpretation. In DLs and their fuzzy extensions, such equivalence relations can be used for concept learning. In this paper, we define and study fuzzy bisimulation and bisimilarity for fuzzy DLs under the Gödel semantics, as well as crisp bisimulation and strong bisimilarity for such logics extended with involutive negation. The considered logics are fuzzy extensions of the DL ALCreg (a variant of PDL) with additional features among inverse roles, nominals, (qualified or unqualified) number restrictions, the universal role, local reflexivity of a role and involutive negation. We formulate and prove results on invariance of concepts under fuzzy (resp. crisp) bisimulation, conditional invariance of fuzzy TBoxex/ABoxes under bisimilarity (resp. strong bisimilarity), and the Hennessy-Milner property of fuzzy (resp. crisp) bisimulation for fuzzy DLs without (resp. with) involutive negation under the Gödel semantics. Apart from these fundamental results, we also provide results on using fuzzy bisimulation to separate the expressive powers of fuzzy DLs, as well as results on using strong bisimilarity to minimize fuzzy interpretations.Linh Anh NguyenQuang Thuy Hathuyhq@vnu.edu.vnNgoc Thanh Nguyenngoc-thanh.nguyen@pwr.wroc.plThi Hong Khanh NguyenThanh Luong Tran2020-10-09T07:11:16Z2020-10-09T07:11:16Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4077This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40772020-10-09T07:11:16ZOn Rectifying the Mapping between Articles and Institutions in Bibliometric DatabasesToday, bibliometric databases are indispensable sources for researchers and research institutions. The main role of these databases is to find research articles and estimate the performance of researchers and institutions. Regarding the evaluation of the research performance of an organization, the accuracy in determining institutions of authors of articles is decisive. However, current popular bibliometric databases such as Scopus and Web of Science have not addressed this point efficiently. To this end, we propose an approach to revise the authors’ affiliation information of articles in bibliometric databases. We build a model to classify articles to institutions with high accuracy by assembling the bag of words and n-grams techniques for extracting features of affiliation strings. After that, these features are weighted to determine their importance to each institution. Affiliation strings of articles are transformed into the new feature space by integrating weights of features and local characteristics of words and phrases contributing to the sequences. Finally, on the feature space, the support vector classifier method is applied to learn a predictive model. Our experimental result shows that the proposed model’s accuracy is about 99.1%.Kien Tuan NgoDinh Hieu Vohieuvd@vnu.edu.vnNgoc Thang Buithangbn@vnu.edu.vnLe Viet Anh PhamKhanh Ly PhamHai Phan2020-10-09T07:10:25Z2020-10-09T07:10:25Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4071This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40712020-10-09T07:10:25ZApplication of WRF-Chem to simulate air quality over Northern VietnamThe WRF-Chem (Weather Research and Forecasting with Chemistry) model is implemented and validated against ground-based observations for meteorological and atmospheric variables for the first time in Northern Vietnam. The WRF-Chem model was based on HTAPv2 emission inventory with MOZCART chemical-aerosol mechanism to simulate atmospheric variables for winter (January) and summer (July) of 2014. The model satisfactorily reproduces meteorological fields, such as temperature 2 m above the ground and relative humidity 2 m above the ground at 45 NCHMF meteorological stations in January, but lower agreement was found in those simulations of July. PM10 and PM2.5 concentrations in January showed good temporal and spatial agreements to observations recorded at three CEM air monitoring stations in Phutho, Quangninh, and Hanoi, with correlation coefficients of 0.36 and 0.59. However, WRF-Chem model was underestimated with MFBs from − 27.9 to − 118.7% for PM10 levels and from − 34.2 to − 115.1% for PM2.5 levels. It has difficulty in capturing day-by-day variation of PM10 and PM2.5 concentrations at each station in July, but MFBs were in the range from − 27.1 to − 40.2% which is slightly lower than those in January. It suggested that further improvements of the model and local emission data are needed to reduce uncertainties in modeling the distribution of atmospheric pollutants. Assessment of biomass burning emission on air quality in summer was analyzed to highlight the application aspect of the WRF-Chem model. The study may serve as a reference for future air quality modeling using WRF-Chem in Vietnam.Thi Nhu Ngoc Dongocdtn@fimo.edu.vnXuan Truong Ngotruongnx@fimo.edu.vnVan Ha Phamhapv@fimo.edu.vnNhu Luan Vuongluannv@cem.gov.vnHoang Anh LeChau Thuy PhamQuang Hung Buihungbq@vnu.edu.vnThi Nhat Thanh Nguyenthanhntn@vnu.edu.vn2020-10-02T03:30:04Z2020-10-02T03:30:04Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4069This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40692020-10-02T03:30:04ZGenerate Test Data from C/C++ Source Code using Weighted CFG and Boundary ValuesThis paper presents two test data automatic generation methods which are based on weighted control flow graph (named WCFT) and boundary values of input parameters (named BVTG). Firstly, WCFT method generates a CFG from a given unit function, updates weight for it, then generates test data from the greatest weight test paths. In the meantime, WCFT can find dead code that can be used for automatic source code errors fix. Secondly, BVTG method generates test data from boundary values of input parameters of the given unit function. The combination of the two generated test data sets from these two methods will improve the error detection ability while maintaining a high code coverage. An implemented tool (named WCFT4Cpp) and experimental results are also presented to show the effectiveness of the two proposed methods in both time required to generate test data and error detection ability.Nguyen Huong Tran17028005@vnu.edu.vnMinh Kha Do17020827@vnu.edu.vnHoang Viet Tranvietth2004@gmail.comNgoc Hung Phamhungpn@vnu.edu.vn2020-09-14T02:55:49Z2020-09-14T02:55:49Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4061This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40612020-09-14T02:55:49ZA new constraint programming model and a linear programming-based adaptive large neighborhood search for the vehicle routing problem with synchronization constraintsWe consider a vehicle routing problem which seeks to minimize cost subject to time window and synchronization constraints. In this problem, the fleet of vehicles is categorized into regular and special vehicles. Some customers require both vehicles’ services, whose service start times at the customer are synchronized. Despite its important real-world application, this problem has rarely been studied in the literature. To solve the problem, we propose a Constraint Programming (CP) model and an Adaptive Large Neighborhood Search (ALNS) in which the design of insertion operators is based on solving linear programming (LP) models to check the insertion feasibility. A number of acceleration techniques is also proposed to significantly reduce the computational time. The computational experiments show that our new CP model finds better solutions than an existing CP-based ALNS, when used on small instances with 25 customers and with a much shorter running time. Our LP-based ALNS dominates the CP-based ALNS, in terms of solution quality, when it provides solutions with better objective values, on average, for all instance classes. This demonstrates the advantage of using linear programming instead of constraint programming when dealing with a variant of vehicle routing problems with relatively tight constraints, which is often considered to be more favorable for CP-based methods. We also adapt our algorithm to solve a well-studied variant of the problem, and the obtained results show that the algorithm provides good solutions as state-of-the-art approaches and improves four best known solutions.Minh Hoang Haminhhoang.ha@vnu.edu.vnTat Dat NguyenDuy Thinh NguyenHoang Giang PhamThuy DoLouis-Martin Rousseau2020-09-14T02:52:10Z2020-09-14T02:52:41Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4060This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40602020-09-14T02:52:10ZSolving the k-dominating set problem on very large-scale networksThe well-known minimum dominating set problem (MDSP) aims to construct the minimum-size subset of vertices in a graph such that every other vertex has at least one neighbor in the subset. In this article, we study a general version of the problem that extends the neighborhood relationship: two vertices are called neighbors of each other if there exists a path through no more than k edges between them. The problem called “minimum k-dominating set problem” (MkDSP) becomes the classical dominating set problem if k is 1 and has important applications in monitoring large-scale social networks. We propose an efficient heuristic algorithm that can handle real-world instances with up to 17 million vertices and 33 million edges. This is the first time such large graphs are solved for the minimum k-dominating set problemMinh Hai NguyenMinh Hoang Haminhhoang.ha@vnu.edu.vnDiep Nguyen NDiep.Nguyen@uts.edu.auThe Trung Trantrung@fpt.edu.vn2020-09-08T09:55:19Z2020-09-08T09:55:19Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4059This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40592020-09-08T09:55:19ZHow to Forecast the Students' Learning Outcomes Based on Factors of Interactive Activities in a Blended Learning CourseThis paper summarizes the research results of identifying the influencing factors in the online learning phase of a blended learning course. From such factors, we propose a model for predicting student outcomes. In our study, we have conducted several models in order to predict the student's learning outcomes, using a course of 231 participants. Obtained data from the logs file of an LMS system is analyzed using learning analytics and machine learning techniques, and the results propose that the four factors are the number of views, the number of posts, the number of forum views, and the number of on-time submitted assignments impact on the student's learning outcomes. For the forecast of the final exam grade based on the results of the formative assessment tests, Bayesian Ridge is the most accurate among the four conducted models (Linear Regression, KNR, SVM, Bayesian Ridge). Our study can be a useful material for lecturers and course designers in effectively organizing blended learning courses.Minh Duc Leduclm@vnu.edu.vnHoa Huy NguyenDuc Loc NguyenViet Anh Nguyenvietanh@vnu.edu.vn2020-08-02T06:23:09Z2020-09-29T11:12:16Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4037This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40372020-08-02T06:23:09ZFormal verification of an abstract version of Anderson protocol with CafeOBJ, CiMPA and CiMPGAnderson protocol is a mutual exclusion protocol. It uses a finite Boolean array shared by all processes and the modulo (or remainder) operation of natural numbers. This is why it is challenging to formally verify that the protocol enjoys the mutual exclusion property in a sense of theorem proving. Then, we make an abstract version of the protocol called A-Anderson protocol that uses an infinite Boolean array instead. We describe how to formally specify A-Anderson protocol in CafeOBJ, an algebraic specification language and how to formally verify that the protocol enjoys the mutual exclusion property in three ways: (1) by writing proof scores in CafeOBJ, (2) with a proof assistant CiMPA for CafeOBJ and (3) with a proof generator CiMPG for CafeOBJ. We mention how to formally verify that Anderson protocol enjoys the property by showing that A-Anderson protocol simulates Anderson protocol.Dinh Duong Tranduongtd@vnu.edu.vnKazuhiro Ogataogata@jaist.ac.jp2020-08-02T06:20:00Z2020-08-02T06:21:31Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4043This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40432020-08-02T06:20:00ZA Lifelong Sentiment Classification Framework Based on a Close Domain Lifelong Topic Modeling MethodIn lifelong machine learning, the determination of the hypotheses related to the current task is very meaningful thanks to the reduction of the space to look for the knowledge patterns supporting for solving the current task. However, there are few studies for this problem. In this paper, we propose the definitions for measuring the “close domains to the current domain”, and a lifelong sentiment classification method based on using the close domains for topic modeling the current domain. Experimental results on sentiment datasets of product reviews from Amazon.com show the promising performance of system and the effectiveness of our approach.Thi Cham Nguyennthicham@hpmu.edu.vnThi Ngan PhamMinh Chau NguyenTri Thanh Nguyenntthanh@vnu.edu.vnQuang Thuy Hathuyhq@vnu.edu.vn2020-08-02T06:19:51Z2020-08-02T06:19:51Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4042This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40422020-08-02T06:19:51ZExploiting CBOW and LSTM Models to Generate Trace Representation for Process MiningIn the field of process mining, one of the challenges of the trace representation problem is to exploit a lot of potentially useful information within the traces while keeping a low dimension of the corresponding vector space. Motivated by the initial results of applying the deep neural networks for producing trace representation, in this paper, we continue to study and apply two more advanced models of deep learning, i.e., Continuous Bag of Words and Long short-term memory, for generating the trace representation. The experimental results have achieved significant improvement, i.e., not only showing the close relationship between the activities in a trace but also helping to reduce the dimension of trace representation.Hong-Nhung BuiTrong-Sinh VuHien-Hanh NguyenTri-Thanh NguyenQuang-Thuy Ha2020-08-02T06:19:24Z2020-08-02T06:19:24Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4038This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40382020-08-02T06:19:24ZMôi trường và giải pháp cho chiến lược quốc gia về trí tuệ nhân tạo của Việt NamTrí tuệ nhân tạo (TTNT) đã tạo ra những chuyển đổi lớn về kinh tế, xã hội, đời sống của con người, và nhiều nước đã xây dựng chiến lược quốc gia về TTNT. Phân tích môi trường và xây dựng chiến lược quốc gia về TTNT là một công việc đầy thách thức đối với nhiều quốc gia, trong đó có Việt Nam. Bài báo tìm hiểu một số nghiên cứu phân tích về môi trường và xây dựng chiến lược quốc gia về TTNT. Từ đó đưa ra giải pháp định hướng cho chiến lược quốc gia về TTNT của Việt NamQuang Thuy Hathuyhq@vnu.edu.vnThanh Thuy Nguyennguyenthanhthuy@vnu.edu.vnBao Son Phamsonpb@vnu.edu.vnXuan Hieu Phanhieupx@vnu.edu.vnTrong Hieu Tranhieutt@vnu.edu.vnMai Vu Tranvutm@vnu.edu.vnQuoc Long Trantqlong@vnu.edu.vnTri Thanh Nguyenntthanh@vnu.edu.vnHoang Tung Ly2020-07-29T08:10:38Z2020-07-29T08:10:38Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4028This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40282020-07-29T08:10:38ZProposing an Elliptic Curve Cryptosystem with the Symmetric
Key for Vietnamese Text Encryption and DecryptionThe article describes the basic idea of Elliptic curve
cryptography (ECC). Elliptic curve arithmetic can be used to
develop Elliptic curve coding schemes, including key exchange,
encryption, and digital signature. The main attraction of Elliptic
curve cryptography compared to RSA is that it provides
equivalent security for a smaller key size, which reduces
processing costs. To encode the Vietnamese text, we are based
on the sound of Vietnamese characters to make a table of these
characters’ order. We are also based on the algorithm to create
the data sequence as the basis of building an encryption
algorithm by using Elliptic curves on finite fields with
symmetric keys to encrypt this Vietnamese textManh Trung MaiPhe Do Ledolp@vnu.edu.vnTrung Thuc Lethuclt12a@gmail.comThi Phuong Anh Dao2020-07-29T08:09:20Z2020-07-29T08:09:20Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3991This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/39912020-07-29T08:09:20ZInformation Security Risk Management by a Holistic Approach: a Case Study for Vietnamese e-GovernmentInformation security risk management is one of the essential tasks currently in ensuring information security. In particular, for e-Government information systems, the assessment and management of security risks through the exploitation of software vulnerabilities, network equipment, etc., allow us to minimize the loss of data and essential information of organizations in e-Government. In this paper, we introduce a holistic approach to assessing information security risks based on both qualitative and quantitative methods for the Vietnamese e-Government. Our model of security risk management is built according to both international standards (ISO 27005-2018, NIST SP800-30r1, SP800-39, SP800-53r4) and Vietnamese standard (TCVN). For the quantitative risk method, we use both CVSS and OWASP scoring standards to quantify information system risks. Besides, the information security risks of the system can also be determined through vulnerability scanners. We also implemented the proposed model in a Web application, called SoC.UET. The experiments we conducted with UET.SoC allowed proving the ability to manage the information security risks holistically for a Ministry or a Province in the Vietnamese e-GovernmenViet Ha Lelevietha@chinhphu.vnVan On Phungphungvanon@gmail.comNgoc Hoa Nguyenhoa.nguyen@vnu.edu.vn2020-07-18T02:00:16Z2020-07-18T02:00:16Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4006This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40062020-07-18T02:00:16ZTowards Blockchainizing Land Valuation Certificate Management Procedures in VietnamNgoc Hoa Nguyenhoa.nguyen@vnu.edu.vnBinh Minh Nguyenminhnb@soict.hust.edu.vnThanh Chung Daochungdt@soict.hust.edu.vnBa Lam Dolamdb@soict.hust.edu.vn2020-07-18T01:59:11Z2020-07-18T01:59:11Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3987This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/39872020-07-18T01:59:11ZIoT Malware Classification Based on System CallsIoT devices play an important role in the industrial revolution 4.0. However, this type of device may exhibit specific security vulnerabilities that can be easily exploited to cause botnet attacks and other malicious activities. In this paper, we introduce a new method for
classification and clustering of IoT malware behaviors
through system call monitoring. Our method is constructed
from multiple one-class SVM classifiers and has the ability to classify known malware with F1-Score over 98% and probability to detect unknown malware up to 97%. Unknown malware instances with similar behaviors can also be grouped together so new classes of malware will be discovered.Dang Kien Hoangkienhd1@vnu.edu.vnDai Tho Nguyennguyendaitho@vnu.edu.vnDuy Loi Vuvuduyloi55@gmail.com2020-07-10T15:05:06Z2020-12-21T09:36:15Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3995This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/39952020-07-10T15:05:06ZSecrecy Performance of Cooperative Cognitive
Radio Networks Under Joint Secrecy Outage and
Primary User Interference ConstraintsIn this paper, we investigate the secrecy performance of a Cooperative Cognitive Radio
Network (CCRN) in the presence of an eavesdropper (EAV). The secondary users (SUs) are subject to three
constraints which include peak transmit power level and interference limitation with respect to the primary
user (PU) as well as secrecy outage constraints due to the EAV. Secrecy outage is achieved when the EAV
cannot decode the targeted signal, but communications in the secondary network is still possible (non-zero
capacity exists). Approximation expressions of the secrecy outage probability and the probability of non-zero
secrecy capacity are derived to evaluate the secrecy performance. Monte Carlo simulations are provided to
examine the accuracy of the derived approximation expressions. Based on this, power allocation policies
for the SUs are derived, satisfying all the constraints while maximizing the secrecy performance as well as
the quality of service performance of the secondary network. It can be concluded that with knowledge of
the channel state information (CSI) of the EAV it is possible to calculate the optimal value for the secrecy
outage threshold of the secondary user (SU) which in turn allows maximizing the secrecy performance. Most
interestingly, our numerical results illustrate that the secrecy performance of the system is much improved
when the parameters obtained using the CSI of the EAV are calculated optimally. Thence, the system can
adjust the power allocation so that no eavesdropping occurs even without reducing quality of service (QoS)
performance compared to a network without any EAV.Xuan Truong Quachqxtruong@ictu.edu.vnHung Trantran.hung@mdh.seElisabeth Uhlemannelisabeth.uhlemann@mdh.seTruc Mai Tranmai.tran@vnu.edu.vn2020-07-10T15:04:40Z2020-07-10T15:04:40Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3990This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/39902020-07-10T15:04:40ZA Fast Computation of the Betweenness Centrality in Social NetworksPhuong Hanh Duhanhdp@vnu.edu.vnNgoc Son Duongduongngocson89@gmail.comNgoc Cuong Nguyencuongnn.hvan@gmail.comNgoc Hoa Nguyenhoa.nguyen@vnu.edu.vn2020-07-10T15:03:43Z2020-07-10T15:03:43Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3989This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/39892020-07-10T15:03:43ZAn effective reversible data hiding method based on pixel-value-orderingThis paper presents a new effective reversible data hiding method based on pixel-value-ordering (iGePVO-K) which is improvement of a recent GePVO-K method that recently is considered as a PVO-used method having highest embedding capacity. In comparison with GePVO-K method, iGePVO-K has the following advantages. First, the embedding capacity of the new method is higher than that of GePVO-K method by using data embedding formulas reasonably and reducing the location map size. Second, for embedding data, in the new method, each pixel value is modified at most by one, while in GePVO-K method, each pixel value may be modified by two. In fact, in the GePVO-K method, the largest pixels are modified by two for embedding bits 1 and by one for bits 0. This is also true for the smallest pixels. Meanwhile, in the proposed method, the largest pixels are modified by one for embedding bits 1 and are unchanged if embedding bits 0. Therefore, the stego-image quality in proposed method is better than that in GePVO-K method. Theoretical analysis and experiment results show that the proposed method has higher embedding capacity and better stego image quality than GePVO-K method.Kim Sao NGUYENsaonkoliver@utc.edu.vnNgoc Hoa Nguyenhoa.nguyen@vnu.edu.vnVan At PHAMphamvanat83@gmail.com2020-07-10T15:00:33Z2020-07-10T15:00:33Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3978This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/39782020-07-10T15:00:33ZA Hybrid Method Based on Genetic Algorithm and Ant Colony System for Traffic Routing OptimizationThis paper presents a hybrid method that combines the genetic algorithm (GA) and the ant colony system algorithm (ACS), namely GACS, to solve the traffic routing problem. In the proposed framework, we use the genetic algorithm to optimize the ACS parameters in order to attain the best trips and travelling time through several novel functions to help ants to update the global and local pheromones. The GACS framework is implemented using the VANETsim package and the real city maps from the open street map project. The experimental results show that our framework achieves a considerably higher performance than A-Star and the classical ACS algorithms in terms of the length of the global best path and the time for trips. Moreover, the GACS framework is also efficient in solving the congestion problem by online monitoring the conditions of traffic light systems.Thi-Hau Nguyennguyenhau@vnu.edu.vnTrung-Tuan DoDuc-Nhan NguyenDang Nhac LuHa Nam Nguyennamnh@vnu.edu.vn2020-07-10T05:51:47Z2020-07-10T05:51:47Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3998This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/39982020-07-10T05:51:47ZAn Efficient Algorithm to Extract Control Flow-based Features for IoT Malware DetectionControl flow-based feature extraction method has the ability to detect malicious code with higher accuracy than traditional text-based methods. Unfortunately, this method has been encountering with the NP-hard problem, which is infeasible for the large-sized and high-complexity programs. To tackle this, we propose a control flow-based features extraction dynamic programming algorithm (CFD) for fast extraction of control flow-based features with polynomial time O(N^2), where N is the number of basic blocks in decompiled executable codes. From the experimental results, it is demonstrated that the proposed algorithm is more efficient and effective in detecting malware than the existing ones. Applying our algorithm to an IoT dataset gives better results on 3 measures: Accuracy (AC) = 99.05%, False Positive Rate (FPR) = 1.31% and False Negative Rate (FNR) = 0.66%.Nghi Phu Trantnphvan@gmail.comDai Tho Nguyennguyendaitho@vnu.edu.vnHuy Hoang Lehoangle.hvan@gmail.comNgoc Toan Nguyenngoctoan.hvan@gmail.comNgoc Binh Nguyennn_binh@kcg.edu2020-07-10T05:40:31Z2020-07-10T05:41:33Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4001This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40012020-07-10T05:40:31ZAutoencoder based friendly jammingPhysical layer security (PLS) provides lightweight security solutions in which security is achieved based on the inherent random characteristics of the wireless medium. In this paper, we consider the PLS approach called friendly jamming (FJ), which is more practical thanks to its low computational complexity. State-of-the-art methods require that legitimate users have full channel state information(CSI)of their channel.Thanks to the recent promising application of the autoencoder (AE) in communication, we propose a new FJ method for PLS using AE without prior knowledge of the CSI. The proposed AE-based FJ method can provide good secrecy performance while avoiding explicit CSI estimation. We also apply the recently proposed tool for mutual information neural estimation (MINE) to evaluate the secrecy capacity. Moreover, we leverage MINE to avoid end-to-end learning in AE-based FJ.Tuan Bui Minhtuanbm.uet@vnu.edu.vnTuyen Ta Ductuyentd@vnu.edu.vnTrung Nguyen Linhlinhtrung@vnu.edu.vnHa Nguyen Viethanv@vnu.edu.vn2020-07-10T05:39:58Z2020-07-10T05:40:05Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/4000This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/40002020-07-10T05:39:58ZCollaborative learning model for cyberattack detection systems in IoT industry 4.0Although the development of IoT Industry 4.0 has brought breakthrough achievements in many sectors, e.g., manufacturing, healthcare, and agriculture, it also raises many security issues to human beings due to a huge of emerging cybersecurity threats recently. In this paper, we propose a novel collaborative learning-based intrusion detection system which can be efficiently implemented in IoT Industry 4.0. In the system under consideration, we develop smart "filters" which can be deployed at the IoT gateways to promptly detect and prevent cyberattacks. In particular, each filter uses the collected data in its network to train its cyberattack detection model based on the deep learning algorithm. After that, the trained model will be shared with other IoT gateways to improve the accuracy in detecting intrusions in the whole system. In this way, not only the detection accuracy is improved, but our proposed system also can significantly reduce the information disclosure as well as network traffic in exchanging data among the IoT gateways. Through thorough simulations on real datasets, we show that the performance obtained by our proposed method can outperform those of the conventional machine learning methods.Khoa Tran Vietkhoatv.uet@vnu.edu.vnSaputra Yuris MulyaYurisMulya.Saputra@student.uts.edu.auHoang Dinh ThaiHoang.Dinh@uts.edu.auTrung Nguyen Linhlinhtrung@vnu.edu.vnDiep Nguyen NDiep.Nguyen@uts.edu.auHa Nguyen Viethanv@vnu.edu.vnDutkiewicz Erykeryk.dutkiewicz@uts.edu.au2020-02-11T08:44:30Z2020-02-11T08:44:57Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3928This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/39282020-02-11T08:44:30ZDynamic Basic Activity Sequence Matching Method in Abnormal Driving Pattern Detection Using Smartphone SensorsIn this work, we present a novel method, namely dynamic basic activity sequence matching (DAS), a combination of machine learning methods and flexible threshold based methods for distinguishing normal and abnormal driving patterns. Indeed, DAS relies on the activity detection module (ADM) presented in our previous work to analyze each driving pattern as a sequence of basic activities—stopping (S), going straight (G), turning left (L), and turning right (R). In fact, the threshold value and other parameters like the duration of long and short activities are iteratively induced from the collected dataset. Hence, DAS is flexible and independent of driving contexts such as vehicle modes and road conditions. Experimental results, on the dataset collected from numerous motorcyclists, show the outperformance of our proposed method against dynamic time warping and the two popular machine learning methods—random forest and neural network—in distinguishing the normal and abnormal driving patterns. Moreover, we propose an efficient framework composing of two phases: in the first phase, the normal and abnormal driving patterns are distinguished by relying on DAS. In the second phase, the detected abnormal patterns are further classified into various specific abnormal driving patterns—weaving, sudden braking, etc. This fusion framework again achieves the highest overall accuracy of 97.94%.Thi Hau Nguyennguyenhau@vnu.edu.vnDang Nhac LuDuc Nhan NguyenHa Nam Nguyennamnh@vnu.edu.vn2020-02-11T08:44:18Z2020-02-11T08:44:18Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3934This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/39342020-02-11T08:44:18ZA Service-based Framework for Building and Executing Epidemic Simulation Applications in the CloudThe cloud has emerged as an attractive platform for resource-intensive scientific applications, such as epidemic simulators. However, building and executing such applications in the cloud presents multiple challenges, including exploiting elasticity, handling failures, and simplifying multi-cloud deployment. To address these challenges, this paper proposes a novel service-based framework called DiFFuSE that enables simulation applications with a bag-of-tasks structure to fully exploit cloud platforms. This paper describes how the framework is applied to restructure two legacy applications, simulating the spread of bovine viral diarrhea virus and Mycobacterium avium subspecies paratuberculosis, into elastic cloud-native applications. Experimental results show that the framework enhances application performance and allows exploring different cost-performance trade-offs while supporting automatic failure handling and elastic resource acquisition from multiple clouds.Nikos Parlavantzasnikos.parlavantzas@irisa.frManh Linh Phamlinhmp@vnu.edu.vnChristine MorinChristine.Morin@inria.frSandie Arnouxsandie.arnoux@inra.frGaël Beaunéegael.beaunee@inra.frLuyuan Qiqiluyuan@gmail.comPhilippe Gontierphilippe.gontier@oniris-nantes.frPauline Ezannopauline.ezanno@oniris-nantes.fr2020-01-31T08:01:34Z2020-01-31T08:01:34Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3923This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/39232020-01-31T08:01:34ZCollaborative Smartphone-Based User Positioning in a Multiple-User Context Using Wireless TechnologiesFor the localization of multiple users, Bluetooth data from the smartphone is able to
complement Wi‐Fi‐based methods with additional information, by providing an approximation of
the relative distances between users. In practice, both positions provided by Wi‐Fi data and relative
distance provided by Bluetooth data are subject to a certain degree of noise due to the uncertainty
of radio propagation in complex indoor environments. In this study, we propose and evaluate two
approaches, namely Non‐temporal and Temporal ones, of collaborative positioning to combine
these two cohabiting technologies to improve the tracking performance. In the Non‐temporal
approach, our model establishes an error observation function in a specific interval of the Bluetooth
and Wi‐Fi output. It is then able to reduce the positioning error by looking for ways to minimize the
error function. The Temporal approach employs an extended error model that takes into account
the time component between users’ movements. For performance evaluation, several multi‐user
scenarios in an indoor environment are set up. Results show that for certain scenarios, the proposed
approaches attain over 40% of improvement in terms of average accuracyViet Cuong Tacuongtv@vnu.edu.vnTrung Kien DaoDominique VaufreydazEric Castelli2019-11-29T15:20:41Z2020-04-25T03:52:58Zhttp://eprints.uet.vnu.edu.vn/eprints/id/eprint/3677This item is in the repository with the URL: http://eprints.uet.vnu.edu.vn/eprints/id/eprint/36772019-11-29T15:20:41ZDomain-independent intent extraction from online textsIdentifying user’s intents from texts on online channels has a wide range of applications from entrepreneurship, banking to e-commerce. However, intent identification is not a simple task due to the intent and its attributes are various and strongly depend on the domain of data. If the number of intent domains increases, the number of intent’s attributes will get bigger. As a result, the complexity of intent extraction task grows up significantly. Additionally, when a new domain comes, it involves considerable physical efforts to define specific labels for intent and attributes for that domain. Hence, it would be much better to come up with a new method for extracting user’s intents which is not dependent on a specific domain.
In our research, we study the problem of domain-independent intent identification from posts and comments crawled from social networks and discussion
forums. We present ten general labels, i.e. labels do not depend on a specific domain, and utilize them when extracting intent and its related information. We also propose a map between general labels and domain-specific labels. We extensively conduct experiments to explore the efficiency of using general labels compared to specific labels in extracting user’s intents when the number of intent domains increases.
Our study is conducted on a medium-sized dataset from three selected domains: Tourism, Real Estate and Transportation. In term of accuracy, when the number of domains grows, our proposal achieves significantly
better results than domain-specific method in identifying user's intent.Thai Le LuongNhu Thuat Tranthuattn@vnu.edu.vnTien Son DangQuoc Long Trantqlong@vnu.edu.vnXuan Hieu Phanhieupx@vnu.edu.vn