Pub Date : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00050
Weiwei Yang, Jie-Si Chen, Yeong-Sheng Chen
This study proposed a management system of the electronic medical records in the blockchain environment. In the proposed system, electronic medical records are first stored in the InterPlanetary File System (IPFS), and then the system generates the hash value, which will be sent to the smart contract to correlate with the patients' data. After that, if the medical staff's medical authority is confirmed by the system, the smart contract can send back the hash value that IPFS generated, and thus the system will present the complete electronic medical records. In the experimental simulation, the results showed that our proposed electronic medical record management and storage process not only can block forged or tampered electronic medical records via smart contracts, but also make it easy to integrate electronic medical records and share medical resources.
{"title":"An Electronic Medical Record Management System Based on Smart Contracts","authors":"Weiwei Yang, Jie-Si Chen, Yeong-Sheng Chen","doi":"10.1109/Ubi-Media.2019.00050","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00050","url":null,"abstract":"This study proposed a management system of the electronic medical records in the blockchain environment. In the proposed system, electronic medical records are first stored in the InterPlanetary File System (IPFS), and then the system generates the hash value, which will be sent to the smart contract to correlate with the patients' data. After that, if the medical staff's medical authority is confirmed by the system, the smart contract can send back the hash value that IPFS generated, and thus the system will present the complete electronic medical records. In the experimental simulation, the results showed that our proposed electronic medical record management and storage process not only can block forged or tampered electronic medical records via smart contracts, but also make it easy to integrate electronic medical records and share medical resources.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121419561","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}
In the mission of natural language processing, sentiment analysis is a formidable challenge due to the complexity of deep network architecture and the lack of standard sentiment word representation. In this paper, we proposed a new learning method of the word representation for the comprehensive information of texts and a minimal Bi-GRU (bidirectional gate recurrent unit) model for the task of sentiment classification. First, for capturing sentiment information of words, the supervised three-layer network is used for construct sentiment word representation. We propose the mixed word representation to denote the classification characteristics, which combines the word embedding of neural probabilistic language model with the proposed the sentiment word representation. Next, we propose bidirectional GRU network including forward and backward propagation to consider the semantic relations before and after sentences, meanwhile, to simple the architecture, we apply minimal GRU network. Then, we combine minimal Bi-GRU model with the mixed word representation taking a full account of semantic and sentiment information to classify the sentiment data set as Movie Reviews and IMDB data set. Experimental results demonstrate that the simplicity of the model and superiority of the performance.
{"title":"Mixed Word Representation and Minimal Bi-GRU Model for Sentiment Analysis","authors":"Yun Liu, Yanping Fu, Yajing Wang, Yong Cui, Zhiyuan Zhang","doi":"10.1109/Ubi-Media.2019.00015","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00015","url":null,"abstract":"In the mission of natural language processing, sentiment analysis is a formidable challenge due to the complexity of deep network architecture and the lack of standard sentiment word representation. In this paper, we proposed a new learning method of the word representation for the comprehensive information of texts and a minimal Bi-GRU (bidirectional gate recurrent unit) model for the task of sentiment classification. First, for capturing sentiment information of words, the supervised three-layer network is used for construct sentiment word representation. We propose the mixed word representation to denote the classification characteristics, which combines the word embedding of neural probabilistic language model with the proposed the sentiment word representation. Next, we propose bidirectional GRU network including forward and backward propagation to consider the semantic relations before and after sentences, meanwhile, to simple the architecture, we apply minimal GRU network. Then, we combine minimal Bi-GRU model with the mixed word representation taking a full account of semantic and sentiment information to classify the sentiment data set as Movie Reviews and IMDB data set. Experimental results demonstrate that the simplicity of the model and superiority of the performance.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125719803","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 : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00012
Yi Ming Chen, C. H. Hsu, Kuo Chung Kuo Chung
Traditional machine learning mostly uses N-gram methods for serialization data prediction, which is not only time-consuming in the pre-processing but also computationally expensive for the model. For the current common malware detection methods, a variety of features such as API, system call, control flow, and permissions are used for machine learning analysis. However, these features depend on expert analysis and to extract multiple features is also time-consuming. This study uses Dalvik opcode as a feature, which is information rich and easy to extract. However, for the time series features of the opcode, the LSTM model and other sequence models will need effective dimension reduction approach because of the long sequence problem and variable feature length, resulting in poor training performance and long training time. Some study uses the training Embedding layer and Autoencoder to reduce the feature dimension. This method requires a layer of network training time. Another method is through feature selection. This method will result in different results as long as the data set changes or the sequence semantic is lost after feature selection. Therefore, in order to solve the above problems, this paper proposes a new preprocessing method to solve the long sequence problem that the LSTM model will encounter, so as to achieve fast training and high accuracy. This study uses a new pre-processing approach combined with an LSTM model to detect malware and achieve 95.58% accuracy on Drebin 10 family and only take 45 seconds to train a model. In addition, in the face of the small training sample problems common to deep learning, this research experiment also proved effective.
{"title":"A Novel Preprocessing Method for Solving Long Sequence Problem in Android Malware Detection","authors":"Yi Ming Chen, C. H. Hsu, Kuo Chung Kuo Chung","doi":"10.1109/Ubi-Media.2019.00012","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00012","url":null,"abstract":"Traditional machine learning mostly uses N-gram methods for serialization data prediction, which is not only time-consuming in the pre-processing but also computationally expensive for the model. For the current common malware detection methods, a variety of features such as API, system call, control flow, and permissions are used for machine learning analysis. However, these features depend on expert analysis and to extract multiple features is also time-consuming. This study uses Dalvik opcode as a feature, which is information rich and easy to extract. However, for the time series features of the opcode, the LSTM model and other sequence models will need effective dimension reduction approach because of the long sequence problem and variable feature length, resulting in poor training performance and long training time. Some study uses the training Embedding layer and Autoencoder to reduce the feature dimension. This method requires a layer of network training time. Another method is through feature selection. This method will result in different results as long as the data set changes or the sequence semantic is lost after feature selection. Therefore, in order to solve the above problems, this paper proposes a new preprocessing method to solve the long sequence problem that the LSTM model will encounter, so as to achieve fast training and high accuracy. This study uses a new pre-processing approach combined with an LSTM model to detect malware and achieve 95.58% accuracy on Drebin 10 family and only take 45 seconds to train a model. In addition, in the face of the small training sample problems common to deep learning, this research experiment also proved effective.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134433410","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 : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00020
Natthapach Anuwattananon, S. Ruengittinun
A gesture from hands and fingers have rich meanings in communication even without a word of sound. It would be very useful if a computer can understand a hand gesture. Hence, we can use a hand gesture to communicate with a robot and perform certain activities. This study focuses on tracking the position of each fingertip and palm to make a computer knows the gesture of a hand. The proposed solution was initially implemented using a MS Kinect camera while capturing a depth image of a human hand. Then, we applied some image processing algorithms to track the positions of fingertips. Finally, the result was visualized in a real-time 3D hand model based on the movements/signs given by a human hand. The experiment results indicate that the proposed approach can literally track the positions of a fingertip.
{"title":"Generating a 3D Hand Model from Position of Fingertip Using Image Processing Technique","authors":"Natthapach Anuwattananon, S. Ruengittinun","doi":"10.1109/Ubi-Media.2019.00020","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00020","url":null,"abstract":"A gesture from hands and fingers have rich meanings in communication even without a word of sound. It would be very useful if a computer can understand a hand gesture. Hence, we can use a hand gesture to communicate with a robot and perform certain activities. This study focuses on tracking the position of each fingertip and palm to make a computer knows the gesture of a hand. The proposed solution was initially implemented using a MS Kinect camera while capturing a depth image of a human hand. Then, we applied some image processing algorithms to track the positions of fingertips. Finally, the result was visualized in a real-time 3D hand model based on the movements/signs given by a human hand. The experiment results indicate that the proposed approach can literally track the positions of a fingertip.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124529810","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 : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00033
Yuan-Tsung Chang, T. Shih
Deep learning is a method that is very commonly used in image recognition. We use the SIFT method to extract feature points, so that the machine can detect the objects in the motion images and they can be integrated into the operation of the robot arm to judge and capture specific objects. This method is also used to detect whether the parameters of the object meet the predetermined values. It will provide a warning if the predetermined values are not met. This can be used to identify the good and defective products on the production line. In the CNN database we have trained more than 30,000 images and improved the last step of SIFT algorithm to demonstrate that our new method can achieve better accuracy.
{"title":"A Deep Learning Approach for Dynamic Object Understanding Using SIFT","authors":"Yuan-Tsung Chang, T. Shih","doi":"10.1109/Ubi-Media.2019.00033","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00033","url":null,"abstract":"Deep learning is a method that is very commonly used in image recognition. We use the SIFT method to extract feature points, so that the machine can detect the objects in the motion images and they can be integrated into the operation of the robot arm to judge and capture specific objects. This method is also used to detect whether the parameters of the object meet the predetermined values. It will provide a warning if the predetermined values are not met. This can be used to identify the good and defective products on the production line. In the CNN database we have trained more than 30,000 images and improved the last step of SIFT algorithm to demonstrate that our new method can achieve better accuracy.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125562793","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}
The expression of Ki-67 with IHC stain has been utilized to assess the prognosis of breast cancer, and the degree of cellular differentiation and proliferation rate. Recently, some researchers utilize the index to predict metastasis of breast carcinoma. In traditional pathological screening, manual assessment of Ki-67 proliferative index may be limited by manual evaluation from different pathologists. Especially, inconsistent biopsy staining would affect the quantitation of Ki-67 proliferation so that developing an automatic system to assess Ki-67 proliferation index poses a big challenge. The goal of this paper is to propose an automatic analysis system to evaluate the degrees of Ki-67 proliferation on IHC stained cells of breast tissue using image processing and machine intelligence techniques. The proposed system not only can assist physicians diagnose, but also provides important information of treatment and prognosis. In order to validate the evaluation performance, we compared with visual assessments by a pathologist and the ImmnuoRatio (i.e., a web-based evaluation system in Ki-67 expression) developed by Vilppu J Tuominen et al.[1] via a number of Ki-67 stained samples for patients with breast carcinoma. Experimental results also demonstrate that the proposed system can automatically, accurately and reliably assess the Ki-67 proliferation index on the breast tissue images with a precision of around 87.37%. However, the accuracy evaluating with ImmunoRatio only can reach 75.82% with the same samples. Moreover, our proposed system also provides various interaction functions including browsing, navigation, and quantitative analyses for pathologists who evaluate the expression of the Ki-67 proliferation.
{"title":"A Whole Slide Ki-67 Proliferation Analysis System for Breast Carcinoma","authors":"C. Ko, Chun-Hung Lin, Chih-Hung Chuang, Chuan-Yu Chang, Shih-Hao Chang, Ji-Han Jiang","doi":"10.1109/Ubi-Media.2019.00048","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00048","url":null,"abstract":"The expression of Ki-67 with IHC stain has been utilized to assess the prognosis of breast cancer, and the degree of cellular differentiation and proliferation rate. Recently, some researchers utilize the index to predict metastasis of breast carcinoma. In traditional pathological screening, manual assessment of Ki-67 proliferative index may be limited by manual evaluation from different pathologists. Especially, inconsistent biopsy staining would affect the quantitation of Ki-67 proliferation so that developing an automatic system to assess Ki-67 proliferation index poses a big challenge. The goal of this paper is to propose an automatic analysis system to evaluate the degrees of Ki-67 proliferation on IHC stained cells of breast tissue using image processing and machine intelligence techniques. The proposed system not only can assist physicians diagnose, but also provides important information of treatment and prognosis. In order to validate the evaluation performance, we compared with visual assessments by a pathologist and the ImmnuoRatio (i.e., a web-based evaluation system in Ki-67 expression) developed by Vilppu J Tuominen et al.[1] via a number of Ki-67 stained samples for patients with breast carcinoma. Experimental results also demonstrate that the proposed system can automatically, accurately and reliably assess the Ki-67 proliferation index on the breast tissue images with a precision of around 87.37%. However, the accuracy evaluating with ImmunoRatio only can reach 75.82% with the same samples. Moreover, our proposed system also provides various interaction functions including browsing, navigation, and quantitative analyses for pathologists who evaluate the expression of the Ki-67 proliferation.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122150273","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 : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00010
L. Yeh, Jiun-Long Huang, Ting-Yin Yen, Jen-Wei Hu
In this paper, we propose a consortium blockchainbased system sharing malicious IP to prevent further attacks happening among other hosts. In our scheme, every security operation center (SOC) serving as a blockchain-node uploads some suspicious IPs to find the potential attackers' IPs. A smart contract is responsible for comparing the loaded IPs and the existing ones without human interference. If IPs in different lists are matched with certain degree, this system will respond by giving the whole list of malicious IP. By means of these steps, shares of IP lists are achieved and attacks are prevented in advance. Besides, when uploading and sharing, we utilize elliptic curve cryptography to ensure data confidentiality and integrity.
{"title":"A Collaborative DDoS Defense Platform Based on Blockchain Technology","authors":"L. Yeh, Jiun-Long Huang, Ting-Yin Yen, Jen-Wei Hu","doi":"10.1109/Ubi-Media.2019.00010","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00010","url":null,"abstract":"In this paper, we propose a consortium blockchainbased system sharing malicious IP to prevent further attacks happening among other hosts. In our scheme, every security operation center (SOC) serving as a blockchain-node uploads some suspicious IPs to find the potential attackers' IPs. A smart contract is responsible for comparing the loaded IPs and the existing ones without human interference. If IPs in different lists are matched with certain degree, this system will respond by giving the whole list of malicious IP. By means of these steps, shares of IP lists are achieved and attacks are prevented in advance. Besides, when uploading and sharing, we utilize elliptic curve cryptography to ensure data confidentiality and integrity.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122681706","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 : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00013
Leo Willyanto Santoso
Nowadays, there is a growing of interest about cloud technology to many companies around the world. That's why many companies trying and implementing cloud computing technologies in their business processes. This research will examine the security requirements that will apply for companies and organizations when they choose to move to a cloud service solution. The study is carried out because cloud services are very desirable in many industries today. Migrating to cloud services would often results in great benefits both financially and administratively. The concerns raised by the transition are how security should be handled. Many companies suffer from a lack of knowledge and it is seen as a big risk to make the transition. This leads to the question that the research strive to answer - which security demands will the transition to a cloud service implicate? In this paper we explain which security requirements are available both for local solutions and cloud solutions. We draw conclusions about what differences there are, what requirements are mutual, which ones are new and which ones are absent if a transition is made to cloud services. The result of this research is an evaluation that companies and organizations can use as a basis when they plan to implement this particular transition.
{"title":"Cloud Technology: Opportunities for Cybercriminals and Security Challenges","authors":"Leo Willyanto Santoso","doi":"10.1109/Ubi-Media.2019.00013","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00013","url":null,"abstract":"Nowadays, there is a growing of interest about cloud technology to many companies around the world. That's why many companies trying and implementing cloud computing technologies in their business processes. This research will examine the security requirements that will apply for companies and organizations when they choose to move to a cloud service solution. The study is carried out because cloud services are very desirable in many industries today. Migrating to cloud services would often results in great benefits both financially and administratively. The concerns raised by the transition are how security should be handled. Many companies suffer from a lack of knowledge and it is seen as a big risk to make the transition. This leads to the question that the research strive to answer - which security demands will the transition to a cloud service implicate? In this paper we explain which security requirements are available both for local solutions and cloud solutions. We draw conclusions about what differences there are, what requirements are mutual, which ones are new and which ones are absent if a transition is made to cloud services. The result of this research is an evaluation that companies and organizations can use as a basis when they plan to implement this particular transition.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128487878","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 : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00051
Wen-Yo Lee, C. Shih, Ti-Hung Chen, Yung-Hui Chen
This paper shows a master-slave module, which is based on a PCB board. The isomorphic circuit board can be the master or the slave through setup the configuration bits. The design can be introduced to several industrial fields, for example, the remote-based control system and the procedure control system, etc. According to the previous work of the researches, the IoT have been introduced to design the industrial equipment; nevertheless, there still have a lot of applications do not have been served. The IoT offers the information all about a system, so the equipment can be taken care anywhere anytime. In this paper, a dental system for the teeth root examining system is developed by the isomorphic master-slave module. It reduces not only the hardware complexity, but also the system development cost.
{"title":"Scalable Master-Slave Isomorphic Module for IoT Service System","authors":"Wen-Yo Lee, C. Shih, Ti-Hung Chen, Yung-Hui Chen","doi":"10.1109/Ubi-Media.2019.00051","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00051","url":null,"abstract":"This paper shows a master-slave module, which is based on a PCB board. The isomorphic circuit board can be the master or the slave through setup the configuration bits. The design can be introduced to several industrial fields, for example, the remote-based control system and the procedure control system, etc. According to the previous work of the researches, the IoT have been introduced to design the industrial equipment; nevertheless, there still have a lot of applications do not have been served. The IoT offers the information all about a system, so the equipment can be taken care anywhere anytime. In this paper, a dental system for the teeth root examining system is developed by the isomorphic master-slave module. It reduces not only the hardware complexity, but also the system development cost.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124383613","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 : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00016
Ning Liu, Bo Shen, Kun Mi, Mingdong Sun, Naiyue Chen
Aspect-based Sentiment analysis (ABSA) is a rapidly growing field of research in natural language processing. ABSA is a fine-grained task of Sentiment analysis. How to capture precise sentiment expressions in a sentence towards the specific aspect remains a challenge. In this paper, we propose a novel neural network, named Multiple-element Attention LSTM (MEA-LSTM) to alleviate the problem of self-attention or binary-element attention used in the ABSA task. These attention mechanisms mentioned above are weak attention, they ignore the information of aspect target or sentence representation. To capture the precise sentiment expressions, we make use of multiple-element attention to assign different importance degrees of different words in a sentence. To store these informative aspect-dependent representations, extra representation memory is designed. Part of speech (POS) is an important feature in identifying the sentiment expressions in the ABSA task. We combine POS with the LSTM in the proposed MEA-LSTM. Experimental results show that our proposed model acquires state-of-the-art accuracy at both restaurant and laptop datasets. Besides, a rule of thumb about choosing the number of hops is given on both datasets.
{"title":"Aspect-Based Sentiment Analysis with the Multiple-Element Attention and Part of Speech","authors":"Ning Liu, Bo Shen, Kun Mi, Mingdong Sun, Naiyue Chen","doi":"10.1109/Ubi-Media.2019.00016","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00016","url":null,"abstract":"Aspect-based Sentiment analysis (ABSA) is a rapidly growing field of research in natural language processing. ABSA is a fine-grained task of Sentiment analysis. How to capture precise sentiment expressions in a sentence towards the specific aspect remains a challenge. In this paper, we propose a novel neural network, named Multiple-element Attention LSTM (MEA-LSTM) to alleviate the problem of self-attention or binary-element attention used in the ABSA task. These attention mechanisms mentioned above are weak attention, they ignore the information of aspect target or sentence representation. To capture the precise sentiment expressions, we make use of multiple-element attention to assign different importance degrees of different words in a sentence. To store these informative aspect-dependent representations, extra representation memory is designed. Part of speech (POS) is an important feature in identifying the sentiment expressions in the ABSA task. We combine POS with the LSTM in the proposed MEA-LSTM. Experimental results show that our proposed model acquires state-of-the-art accuracy at both restaurant and laptop datasets. Besides, a rule of thumb about choosing the number of hops is given on both datasets.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133407259","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：