Pub Date : 2022-03-20DOI: 10.15625/1813-9663/38/1/16720
Tran Manh Tuan, Phung The Huan, Pham Huy Thong, T. Ngan, Le Hoang Son
Data clustering are applied in various fields such as document classification, dental X-ray image segmentation, medical image segmentation, etc. Especially, clustering algorithms are used in satellite image processing in many important application areas, including classification of vehicles participating in traffic, logistics, classification of satellite images to forecast droughts, floods, forest fire, etc. In the process of collecting satellite image data, there are a number of factors such as clouds, weather, ... that can affect to image quality. Images with low quality will make the performance of clustering algorithms decrease. Apart from that, the parameter of fuzzification in clustering algorithms also affects to clustering results. In the past, clustering methods often used the same fuzzification parameter, m = 2. But in practice, each element should have its own parameter m. Therefore, determining the parameters m is necessary to increase fuzzy clustering performance. In this research, an improvement algorithm for the data partition with confidence problem and multi fuzzifier named as TS3MFCM is introduced. The proposed method consists of three steps namely as “FCM for labeled data”, “Data transformation”, and “Semi-supervised fuzzy clustering with multiple point fuzzifiers”. The proposed TS3MFCM method is implemented and experimentally compared against with the Confidence-weighted Safe Semi-Supervised Clustering (CS3FCM). The performance of proposed method is better than selected methods in both computational time and clustering accuracy on the same datasets
{"title":"AN IMPROVEMENT OF TRUSTED SAFE SEMI-SUPERVISED FUZZY CLUSTERING METHOD WITH MULTIPLE FUZZIFIERS","authors":"Tran Manh Tuan, Phung The Huan, Pham Huy Thong, T. Ngan, Le Hoang Son","doi":"10.15625/1813-9663/38/1/16720","DOIUrl":"https://doi.org/10.15625/1813-9663/38/1/16720","url":null,"abstract":"Data clustering are applied in various fields such as document classification, dental X-ray image segmentation, medical image segmentation, etc. Especially, clustering algorithms are used in satellite image processing in many important application areas, including classification of vehicles participating in traffic, logistics, classification of satellite images to forecast droughts, floods, forest fire, etc. In the process of collecting satellite image data, there are a number of factors such as clouds, weather, ... that can affect to image quality. Images with low quality will make the performance of clustering algorithms decrease. Apart from that, the parameter of fuzzification in clustering algorithms also affects to clustering results. In the past, clustering methods often used the same fuzzification parameter, m = 2. But in practice, each element should have its own parameter m. Therefore, determining the parameters m is necessary to increase fuzzy clustering performance. In this research, an improvement algorithm for the data partition with confidence problem and multi fuzzifier named as TS3MFCM is introduced. The proposed method consists of three steps namely as “FCM for labeled data”, “Data transformation”, and “Semi-supervised fuzzy clustering with multiple point fuzzifiers”. The proposed TS3MFCM method is implemented and experimentally compared against with the Confidence-weighted Safe Semi-Supervised Clustering (CS3FCM). The performance of proposed method is better than selected methods in both computational time and clustering accuracy on the same datasets","PeriodicalId":15444,"journal":{"name":"Journal of Computer Science and Cybernetics","volume":"6 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2022-03-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88789082","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-20DOI: 10.15625/1813-9663/38/1/16537
Nguyen Hien Trinh, Doan Van Ban, Vu Vinh Quang, Cáp Thanh Tùng
Detecting community structure on social network has been an important and interesting issue on which many researchers have paid much attention and developed applications. Many graph clustering algorithms have been applied to find disjoint communities, i.e each node belongs to a single community. However, for social network in particular, public communication network in general, most of communities are not completely detached but they may be embedding, overlapping or crossing, that means certain nodes can belong to more than one community. Overlapping node plays a role of interface between communities and it is really interesting to study the community establishment of these nodes because it reflects dynamic behaviuor of participants.This article introduces the algorithm to find overlapping communities on huge social network. The proposed COPACN algorithm has been developed on the basis of label propagation, using advanced clustering coefficient to find overlapping communities on social network. Exprermental results on a set of popular, standard social networks and certain real network have shown the high speed and high effiency in finding overlapping structures.
{"title":"A FAST OVERLAPPING COMMUNITY DETECTION ALGORITHM BASED ON LABEL PROPAGATION AND SOCIAL NETWORK GRAPH CLUSTERING COEFFICIENT","authors":"Nguyen Hien Trinh, Doan Van Ban, Vu Vinh Quang, Cáp Thanh Tùng","doi":"10.15625/1813-9663/38/1/16537","DOIUrl":"https://doi.org/10.15625/1813-9663/38/1/16537","url":null,"abstract":"Detecting community structure on social network has been an important and interesting issue on which many researchers have paid much attention and developed applications. Many graph clustering algorithms have been applied to find disjoint communities, i.e each node belongs to a single community. However, for social network in particular, public communication network in general, most of communities are not completely detached but they may be embedding, overlapping or crossing, that means certain nodes can belong to more than one community. Overlapping node plays a role of interface between communities and it is really interesting to study the community establishment of these nodes because it reflects dynamic behaviuor of participants.This article introduces the algorithm to find overlapping communities on huge social network. The proposed COPACN algorithm has been developed on the basis of label propagation, using advanced clustering coefficient to find overlapping communities on social network. Exprermental results on a set of popular, standard social networks and certain real network have shown the high speed and high effiency in finding overlapping structures.","PeriodicalId":15444,"journal":{"name":"Journal of Computer Science and Cybernetics","volume":"41 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2022-03-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73442782","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-20DOI: 10.15625/1813-9663/38/1/16309
Trinh Thi Loan, P. T. Anh, Le Viet Nam, Hoang Van Dung
This paper presents a deep learning model to address the problem of recognition of animals and plants. The context of this work is to make an effort in protection of rare species that are seriously faced to the risk of extinction in Vietnam such as Panthera pardus, Dalbergia cochinchinensis, Macaca mulatta. The proposed approach exploits the advanced learning ability of convolutional neural networks and Inception residual structures to design a lightweight model for classification task. We also apply the transfer learning technique to fine-tune the two state-of-the-art methods, MobileNetV2 and InceptionV3, specific to our own dataset. Experimental results demonstrate the superiority of our object predictor (e.g., 95.8% accuracy) in comparison with other methods. In addition, the proposed model works very efficiently with the inference speed of around 113 FPS on a CPU machine, enabling it for deployment on mobile environment.
{"title":"AN EFFECTIVE DEEP LEARNING MODEL FOR RECOGNITION OF ANIMALS AND PLANTS","authors":"Trinh Thi Loan, P. T. Anh, Le Viet Nam, Hoang Van Dung","doi":"10.15625/1813-9663/38/1/16309","DOIUrl":"https://doi.org/10.15625/1813-9663/38/1/16309","url":null,"abstract":"This paper presents a deep learning model to address the problem of recognition of animals and plants. The context of this work is to make an effort in protection of rare species that are seriously faced to the risk of extinction in Vietnam such as Panthera pardus, Dalbergia cochinchinensis, Macaca mulatta. The proposed approach exploits the advanced learning ability of convolutional neural networks and Inception residual structures to design a lightweight model for classification task. We also apply the transfer learning technique to fine-tune the two state-of-the-art methods, MobileNetV2 and InceptionV3, specific to our own dataset. Experimental results demonstrate the superiority of our object predictor (e.g., 95.8% accuracy) in comparison with other methods. In addition, the proposed model works very efficiently with the inference speed of around 113 FPS on a CPU machine, enabling it for deployment on mobile environment.","PeriodicalId":15444,"journal":{"name":"Journal of Computer Science and Cybernetics","volume":"124 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2022-03-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88020711","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-20DOI: 10.15625/1813-9663/38/1/16092
Tam Thanh Nguyen, Toan Thanh Nguyen, C. T. Phan, Quoc Viet Hung Nguyen
Changes in surface water might result in natural disasters such as floods, water shortages, landslides, waterborne diseases, which lead to loss of lives. Timely extracting for surface water and predicting its movement is essential for planning activities and decision-making processes. Most existing works on extracting water surface using satellite images focus on static spectral images and ignore the temporal evolution of data in streams, leading to less accuracy and lack of prediction power. Although some works realize that modeling temporal information of satellite signals could boost the forecasting capability on environmental changes, most of them only focus on prediction tasks independently and separately from the extraction task. In this paper, we propose a unified framework for water extraction and change prediction (WECP) built on top of imagery data streams, which are free to access from orbiting satellites, to locate water surface and predict its changes over time. Our framework is evaluated on Landsat 8 data due to its high spatial resolution. Empirical evaluations on real imagery datasets of different landscapes reveal that our framework is robust in extracting and capturing spatio-temporal changes in the water surface.
{"title":"A UNIFIED FRAMEWORK FOR WATER SURFACE EXTRACTION AND CHANGE PREDICTION IN IMAGERY DATA STREAMS","authors":"Tam Thanh Nguyen, Toan Thanh Nguyen, C. T. Phan, Quoc Viet Hung Nguyen","doi":"10.15625/1813-9663/38/1/16092","DOIUrl":"https://doi.org/10.15625/1813-9663/38/1/16092","url":null,"abstract":"Changes in surface water might result in natural disasters such as floods, water shortages, landslides, waterborne diseases, which lead to loss of lives. Timely extracting for surface water and predicting its movement is essential for planning activities and decision-making processes. Most existing works on extracting water surface using satellite images focus on static spectral images and ignore the temporal evolution of data in streams, leading to less accuracy and lack of prediction power. Although some works realize that modeling temporal information of satellite signals could boost the forecasting capability on environmental changes, most of them only focus on prediction tasks independently and separately from the extraction task. In this paper, we propose a unified framework for water extraction and change prediction (WECP) built on top of imagery data streams, which are free to access from orbiting satellites, to locate water surface and predict its changes over time. Our framework is evaluated on Landsat 8 data due to its high spatial resolution. Empirical evaluations on real imagery datasets of different landscapes reveal that our framework is robust in extracting and capturing spatio-temporal changes in the water surface.","PeriodicalId":15444,"journal":{"name":"Journal of Computer Science and Cybernetics","volume":"22 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2022-03-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75048788","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-20DOI: 10.15625/1813-9663/38/1/15961
Chu Ba Thanh, Trinh Van Loan, Dao Thi Dieu Thuy
We can say that music in general is an indispensable spiritual food in human life. For Vietnamese people, folk music plays a very important role, it has entered the minds of every Vietnamese person right from the moment of birth through lullabies for children. In Vietnam, there are many different types of folk songs that everyone loves, and each has many different melodies. In order to archive and search music works with a very large quantity, including folk songs, it is necessary to automatically classify and identify those works. This paper presents the method of determining the feature parameters and then using the convolution neural network (CNN) to classify and identify some Vietnamese folk tunes as Quanho and Cheo. Our experimental results show that the average highest classification and identification accuracy are 99.92% and 97.67%, respectivel.
{"title":"AUTOMATIC IDENTIFICATION OF SOME VIETNAMESE FOLK SONGS CHEO AND QUANHO USING CONVOLUTIONAL NEURAL NETWORKS","authors":"Chu Ba Thanh, Trinh Van Loan, Dao Thi Dieu Thuy","doi":"10.15625/1813-9663/38/1/15961","DOIUrl":"https://doi.org/10.15625/1813-9663/38/1/15961","url":null,"abstract":"We can say that music in general is an indispensable spiritual food in human life. For Vietnamese people, folk music plays a very important role, it has entered the minds of every Vietnamese person right from the moment of birth through lullabies for children. In Vietnam, there are many different types of folk songs that everyone loves, and each has many different melodies. In order to archive and search music works with a very large quantity, including folk songs, it is necessary to automatically classify and identify those works. This paper presents the method of determining the feature parameters and then using the convolution neural network (CNN) to classify and identify some Vietnamese folk tunes as Quanho and Cheo. Our experimental results show that the average highest classification and identification accuracy are 99.92% and 97.67%, respectivel.","PeriodicalId":15444,"journal":{"name":"Journal of Computer Science and Cybernetics","volume":"118 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2022-03-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87649537","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-12DOI: 10.15625/1813-9663/37/4/16180
Quang Minh Ha, Duy Manh Vu, Xuan Thanh Le, M. Hoang
This paper deals with the Traveling Salesman Problem with Multi-Visit Drone (TSP-MVD) in which a truck works in collaboration with a drone that can serve up to q > 1 customers consecutively during each sortie. We propose a Mixed Integer Linear Programming (MILP) formulation and a metaheuristic based on Iterated Local Search to solve the problem. Benchmark instances collected from the literature of the special case with q = 1 are used to test the performance of our algorithms. The obtained results show that our MILP model can solve a number of instances to optimality. This is the first time optimal solutions for these instances are reported. Our ILS performs better other algorithms in terms of both solution quality and running time on several class of instances. The numerical results obtained by testing the methods on new randomly generated instances show again the effectiveness of the methods as well as the positive impact of using the multi-visit drone.
{"title":"THE TRAVELING SALESMAN PROBLEM WITH MULTI-VISIT DRONE","authors":"Quang Minh Ha, Duy Manh Vu, Xuan Thanh Le, M. Hoang","doi":"10.15625/1813-9663/37/4/16180","DOIUrl":"https://doi.org/10.15625/1813-9663/37/4/16180","url":null,"abstract":"This paper deals with the Traveling Salesman Problem with Multi-Visit Drone (TSP-MVD) in which a truck works in collaboration with a drone that can serve up to q > 1 customers consecutively during each sortie. We propose a Mixed Integer Linear Programming (MILP) formulation and a metaheuristic based on Iterated Local Search to solve the problem. Benchmark instances collected from the literature of the special case with q = 1 are used to test the performance of our algorithms. The obtained results show that our MILP model can solve a number of instances to optimality. This is the first time optimal solutions for these instances are reported. Our ILS performs better other algorithms in terms of both solution quality and running time on several class of instances. The numerical results obtained by testing the methods on new randomly generated instances show again the effectiveness of the methods as well as the positive impact of using the multi-visit drone.","PeriodicalId":15444,"journal":{"name":"Journal of Computer Science and Cybernetics","volume":"61 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85336297","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-12DOI: 10.15625/1813-9663/37/4/15892
H. Phan, N. Nguyen, D. Hwang
With the rapid development of the Internet industry, an increasing number of social media platforms have been developed. These social media platforms have become the main channels for communication among most users. Opinions from social media platforms provide the most updated and inclusive information. Sentiments from opinions are a valuable data source for solving many issues. Therefore, sentiment analysis has developed into one of the most popular natural language processing fields. Hence, improving the performance of sentiment analysis methods or discovering new problems related to these methods is essential. In this context, we must be aware of the general information relevant to this area. This survey presents a summary of the necessary stages for building a complete model to be used in sentiment analysis. For each procedure, we list the popular techniques that have been widely used in recent years. In addition, discussions and comparisons related to these methods are provided. Additionally, we discuss the challenges and possible research directions for future research in this field.
{"title":"SENTIMENT ANALYSIS FOR SOCIAL MEDIA: A SURVEY","authors":"H. Phan, N. Nguyen, D. Hwang","doi":"10.15625/1813-9663/37/4/15892","DOIUrl":"https://doi.org/10.15625/1813-9663/37/4/15892","url":null,"abstract":"With the rapid development of the Internet industry, an increasing number of social media platforms have been developed. These social media platforms have become the main channels for communication among most users. Opinions from social media platforms provide the most updated and inclusive information. Sentiments from opinions are a valuable data source for solving many issues. Therefore, sentiment analysis has developed into one of the most popular natural language processing fields. Hence, improving the performance of sentiment analysis methods or discovering new problems related to these methods is essential. In this context, we must be aware of the general information relevant to this area. This survey presents a summary of the necessary stages for building a complete model to be used in sentiment analysis. For each procedure, we list the popular techniques that have been widely used in recent years. In addition, discussions and comparisons related to these methods are provided. Additionally, we discuss the challenges and possible research directions for future research in this field.","PeriodicalId":15444,"journal":{"name":"Journal of Computer Science and Cybernetics","volume":"3 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90346805","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-26DOI: 10.15625/1813-9663/37/3/16106
N. C. Ho
The study stands on the standpoint that there exist relationships between real-world structures and their provided information in reality. Such relationships are essential because the natural language plays a specifically vital and crucial role in, e.g., capturing, conveying information, and accumulating knowledge containing useful high-level information. Consequently, it must contain certain semantics structures, including linguistic (L-) variables’ semantic structures, which are fundamental, similar to the math variables’ structures. In this context, the fact that the (L-) variables’ word domains can be formalized as algebraic semantics-based structures in an axiomatic manner, called hedge algebras (HAs,) is still a novel event and essential for developing computational methods to simulate the human capabilities in problem-solving based on the so-called natural language-based formalism. Hedge algebras were founded in 1990. Since then, HA-formalism has been significantly developed and applied to solve several application problems in many distinct fields, such as fuzzy control, data classification and regression, robotics, L-time series forecasting, and L-data summarization. The study gives a survey to summarize specific distinguishing fundamental features of HA-formalism, its applicability in problem-solving, and its performance.
{"title":"INFORMATION AND MATHEMATICAL STRUCTURES CONTAINED IN THE NATURAL LANGUAGE WORD DOMAINS AND THEIR APPLICATIONS","authors":"N. C. Ho","doi":"10.15625/1813-9663/37/3/16106","DOIUrl":"https://doi.org/10.15625/1813-9663/37/3/16106","url":null,"abstract":"The study stands on the standpoint that there exist relationships between real-world structures and their provided information in reality. Such relationships are essential because the natural language plays a specifically vital and crucial role in, e.g., capturing, conveying information, and accumulating knowledge containing useful high-level information. Consequently, it must contain certain semantics structures, including linguistic (L-) variables’ semantic structures, which are fundamental, similar to the math variables’ structures. In this context, the fact that the (L-) variables’ word domains can be formalized as algebraic semantics-based structures in an axiomatic manner, called hedge algebras (HAs,) is still a novel event and essential for developing computational methods to simulate the human capabilities in problem-solving based on the so-called natural language-based formalism. Hedge algebras were founded in 1990. Since then, HA-formalism has been significantly developed and applied to solve several application problems in many distinct fields, such as fuzzy control, data classification and regression, robotics, L-time series forecasting, and L-data summarization. The study gives a survey to summarize specific distinguishing fundamental features of HA-formalism, its applicability in problem-solving, and its performance. ","PeriodicalId":15444,"journal":{"name":"Journal of Computer Science and Cybernetics","volume":"4 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85646114","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-26DOI: 10.15625/1813-9663/37/3/15962
A. Aharari, J. Abe, K. Nakamatsu
The coronavirus (COVID-19) is the latest pandemic that hit human health in 2019. Wear a face mask in public areas to decrease the spread of the coronavirus. This work presents real-time face mask detection with facial temperature measures for the COVID-19 indoor monitoring system. Detecting people using ultrasonic sensors, face mask detection, and facial temperature measure using Grid-Eye Sensor are three modules applied in the proposed system. We also evaluated the proposed monitoring system in the real environment and confirmed the accuracy of 98.8% of mask detection.
{"title":"DEVELOPING REAL-TIME FACE MASK DETECTION WITH FACIAL TEMPERATURE MEASURE FOR COVID-19 INDOOR MONITORING SYSTEM","authors":"A. Aharari, J. Abe, K. Nakamatsu","doi":"10.15625/1813-9663/37/3/15962","DOIUrl":"https://doi.org/10.15625/1813-9663/37/3/15962","url":null,"abstract":"The coronavirus (COVID-19) is the latest pandemic that hit human health in 2019. Wear a face mask in public areas to decrease the spread of the coronavirus. This work presents real-time face mask detection with facial temperature measures for the COVID-19 indoor monitoring system. Detecting people using ultrasonic sensors, face mask detection, and facial temperature measure using Grid-Eye Sensor are three modules applied in the proposed system. We also evaluated the proposed monitoring system in the real environment and confirmed the accuracy of 98.8% of mask detection.","PeriodicalId":15444,"journal":{"name":"Journal of Computer Science and Cybernetics","volume":"22 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76523201","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-05-31DOI: 10.15625/1813-9663/37/2/15980
Tuan Minh Luu, H. T. Le, T. Hoang
Deep neural networks have been applied successfully to extractive text summarization tasks with the accompany of large training datasets. However, when the training dataset is not large enough, these models reveal certain limitations that affect the quality of the system’s summary. In this paper, we propose an extractive summarization system basing on a Convolutional Neural Network and a Fully Connected network for sentence selection. The pretrained BERT multilingual model is used to generate embeddings vectors from the input text. These vectors are combined with TF-IDF values to produce the input of the text summarization system. Redundant sentences from the output summary are eliminated by the Maximal Marginal Relevance method. Our system is evaluated with both English and Vietnamese languages using CNN and Baomoi datasets, respectively. Experimental results show that our system achieves better results comparing to existing works using the same dataset. It confirms that our approach can be effectively applied to summarize both English and Vietnamese languages.
{"title":"A HYBRID MODEL USING THE PRETRAINED BERT AND DEEP NEURAL NETWORKS WITH RICH FEATURE FOR EXTRACTIVE TEXT SUMMARIZATION","authors":"Tuan Minh Luu, H. T. Le, T. Hoang","doi":"10.15625/1813-9663/37/2/15980","DOIUrl":"https://doi.org/10.15625/1813-9663/37/2/15980","url":null,"abstract":"Deep neural networks have been applied successfully to extractive text summarization tasks with the accompany of large training datasets. However, when the training dataset is not large enough, these models reveal certain limitations that affect the quality of the system’s summary. In this paper, we propose an extractive summarization system basing on a Convolutional Neural Network and a Fully Connected network for sentence selection. The pretrained BERT multilingual model is used to generate embeddings vectors from the input text. These vectors are combined with TF-IDF values to produce the input of the text summarization system. Redundant sentences from the output summary are eliminated by the Maximal Marginal Relevance method. Our system is evaluated with both English and Vietnamese languages using CNN and Baomoi datasets, respectively. Experimental results show that our system achieves better results comparing to existing works using the same dataset. It confirms that our approach can be effectively applied to summarize both English and Vietnamese languages.","PeriodicalId":15444,"journal":{"name":"Journal of Computer Science and Cybernetics","volume":"25 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-05-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85756918","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：