Pub Date : 2021-03-03DOI: 10.1109/CSICC52343.2021.9420580
F. Manavi, A. Hamzeh
Ransomware is a type of malware from cryptovirology that perpetually blocks access to a victim’s data unless a ransom is paid. Today, this type of malware has grown dramatically and has targeted the computer systems of some important organizations such as hospitals, banks, and Water Organization. Therefore, early detection of this type of malware is very important. This paper describes a solution to ransomware detection based on executable file headers. Header of the executable file expresses important information about the structure of the program. In other words, the header’s information is a sequence of bytes, and changing it changes the structure of the program file. In the proposed method, using LSTM network, the sequence of bytes that constructs the header is processed and the ransomware samples are separated from the benign samples. The proposed method can detect a ransomware sample with 93.25 accuracy without running the program and using a raw header, so it is suitable for quick detection of suspicious samples.
{"title":"Static Detection of Ransomware Using LSTM Network and PE Header","authors":"F. Manavi, A. Hamzeh","doi":"10.1109/CSICC52343.2021.9420580","DOIUrl":"https://doi.org/10.1109/CSICC52343.2021.9420580","url":null,"abstract":"Ransomware is a type of malware from cryptovirology that perpetually blocks access to a victim’s data unless a ransom is paid. Today, this type of malware has grown dramatically and has targeted the computer systems of some important organizations such as hospitals, banks, and Water Organization. Therefore, early detection of this type of malware is very important. This paper describes a solution to ransomware detection based on executable file headers. Header of the executable file expresses important information about the structure of the program. In other words, the header’s information is a sequence of bytes, and changing it changes the structure of the program file. In the proposed method, using LSTM network, the sequence of bytes that constructs the header is processed and the ransomware samples are separated from the benign samples. The proposed method can detect a ransomware sample with 93.25 accuracy without running the program and using a raw header, so it is suitable for quick detection of suspicious samples.","PeriodicalId":374593,"journal":{"name":"2021 26th International Computer Conference, Computer Society of Iran (CSICC)","volume":"77 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123833190","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 : 2021-03-03DOI: 10.1109/CSICC52343.2021.9420546
Mohamadreza Bakhtyari, S. Mirzaei
Click-Through Rate (CTR) prediction plays a critical role in online advertisement campaigns and recommendation systems. Most of the state-of-the-art models are based on Factorization Machines and some of these models try to feed mapped field features to a deep learning component for learning users’ interests by modelling feature interactions. Deploying a model for CTR is an online task and should be able to perform well with a limited amount of data and time. While these models are very good at prediction inferences and learning feature interactions, their deep component needs a vast amount of data and time and does not perform well in limited situations.In a recent article, a combination of boosting algorithms with deep factorization machines (XDBoost algorithm) has been proposed. In this paper, we use a boosting algorithm for prediction inference with limited raw data and time. We show that with an appropriate feature engineering and fine parameter tuning for a raw boosting model, we can outperform XDBoost method and get better results. We will use exploratory data analysis to extract the main characteristics of the dataset and eliminate the redundant data. Then, by applying grid search scheme, we select the best values for the hyperparameters of our model.
{"title":"Click-Through Rate Prediction Using Feature Engineered Boosting Algorithms","authors":"Mohamadreza Bakhtyari, S. Mirzaei","doi":"10.1109/CSICC52343.2021.9420546","DOIUrl":"https://doi.org/10.1109/CSICC52343.2021.9420546","url":null,"abstract":"Click-Through Rate (CTR) prediction plays a critical role in online advertisement campaigns and recommendation systems. Most of the state-of-the-art models are based on Factorization Machines and some of these models try to feed mapped field features to a deep learning component for learning users’ interests by modelling feature interactions. Deploying a model for CTR is an online task and should be able to perform well with a limited amount of data and time. While these models are very good at prediction inferences and learning feature interactions, their deep component needs a vast amount of data and time and does not perform well in limited situations.In a recent article, a combination of boosting algorithms with deep factorization machines (XDBoost algorithm) has been proposed. In this paper, we use a boosting algorithm for prediction inference with limited raw data and time. We show that with an appropriate feature engineering and fine parameter tuning for a raw boosting model, we can outperform XDBoost method and get better results. We will use exploratory data analysis to extract the main characteristics of the dataset and eliminate the redundant data. Then, by applying grid search scheme, we select the best values for the hyperparameters of our model.","PeriodicalId":374593,"journal":{"name":"2021 26th International Computer Conference, Computer Society of Iran (CSICC)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134497037","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 : 2021-03-03DOI: 10.1109/CSICC52343.2021.9420603
M. Imani
A polarimetric synthetic aperture radar (PolSAR) image classification is introduced in this work. The proposed method called as ridge regression-based polarimetric-spatial (RRPS) feature extraction generates polarimetric-spatial features with minimum overlapping and redundant information. To this end, each polarimetric-spatial channel of PolSAR data is represented through a ridge regression model using the farthest neighbors of that channel. The weights of the regression model compose the projection matrix for dimensionality reduction. The proposed RRPS method with a closed form solution has high performance in PolSAR image classification using small training sets.
{"title":"Polarimetric SAR Classification Using Ridge Regression-Based Polarimetric-Spatial Feature Extraction","authors":"M. Imani","doi":"10.1109/CSICC52343.2021.9420603","DOIUrl":"https://doi.org/10.1109/CSICC52343.2021.9420603","url":null,"abstract":"A polarimetric synthetic aperture radar (PolSAR) image classification is introduced in this work. The proposed method called as ridge regression-based polarimetric-spatial (RRPS) feature extraction generates polarimetric-spatial features with minimum overlapping and redundant information. To this end, each polarimetric-spatial channel of PolSAR data is represented through a ridge regression model using the farthest neighbors of that channel. The weights of the regression model compose the projection matrix for dimensionality reduction. The proposed RRPS method with a closed form solution has high performance in PolSAR image classification using small training sets.","PeriodicalId":374593,"journal":{"name":"2021 26th International Computer Conference, Computer Society of Iran (CSICC)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125553091","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 : 2021-03-03DOI: 10.1109/CSICC52343.2021.9420549
Razieh Baradaran, Hossein Amirkhani
One of the main challenges of the machine reading comprehension (MRC) models is their fragile out-of-domain generalization, which makes these models not properly applicable to real-world general-purpose question answering problems. In this paper, we leverage a zero-shot weighted ensemble method for improving the robustness of out-of-domain generalization in MRC models. In the proposed method, a weight estimation module is used to estimate out-of-domain weights, and an ensemble module aggregate several base models’ predictions based on their weights. The experiments indicate that the proposed method not only improves the final accuracy, but also is robust against domain changes.
{"title":"Zero-Shot Estimation of Base Models’ Weights in Ensemble of Machine Reading Comprehension Systems for Robust Generalization","authors":"Razieh Baradaran, Hossein Amirkhani","doi":"10.1109/CSICC52343.2021.9420549","DOIUrl":"https://doi.org/10.1109/CSICC52343.2021.9420549","url":null,"abstract":"One of the main challenges of the machine reading comprehension (MRC) models is their fragile out-of-domain generalization, which makes these models not properly applicable to real-world general-purpose question answering problems. In this paper, we leverage a zero-shot weighted ensemble method for improving the robustness of out-of-domain generalization in MRC models. In the proposed method, a weight estimation module is used to estimate out-of-domain weights, and an ensemble module aggregate several base models’ predictions based on their weights. The experiments indicate that the proposed method not only improves the final accuracy, but also is robust against domain changes.","PeriodicalId":374593,"journal":{"name":"2021 26th International Computer Conference, Computer Society of Iran (CSICC)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125207748","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 : 2021-03-03DOI: 10.1109/CSICC52343.2021.9420619
F. Farahani, F. Rezaei
The Internet of things (IoT) is generating a huge amount of data and big data management is of key importance. One of the important applications of IoT is smart meter networks and one of the key issues in establishing smart meter networks is managing the large volume of data sent by the meters. In this paper, we present a data management system implemented for monitoring and managing the data collected from the smart meters and controlling them in a large-scale network. IoT infrastructure with LPWAN (Low Power Wide Area Network) class is considered in this system. Moreover, two methods are proposed to improve the performance in terms of scalability and response time. It is shown that the implemented data management system using the proposed methods achieves significant performance improvement in large scale networks.
{"title":"Implementing a Scalable Data Management System for Collected Data by Smart Meters","authors":"F. Farahani, F. Rezaei","doi":"10.1109/CSICC52343.2021.9420619","DOIUrl":"https://doi.org/10.1109/CSICC52343.2021.9420619","url":null,"abstract":"The Internet of things (IoT) is generating a huge amount of data and big data management is of key importance. One of the important applications of IoT is smart meter networks and one of the key issues in establishing smart meter networks is managing the large volume of data sent by the meters. In this paper, we present a data management system implemented for monitoring and managing the data collected from the smart meters and controlling them in a large-scale network. IoT infrastructure with LPWAN (Low Power Wide Area Network) class is considered in this system. Moreover, two methods are proposed to improve the performance in terms of scalability and response time. It is shown that the implemented data management system using the proposed methods achieves significant performance improvement in large scale networks.","PeriodicalId":374593,"journal":{"name":"2021 26th International Computer Conference, Computer Society of Iran (CSICC)","volume":"496 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122783190","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 : 2021-03-03DOI: 10.1109/CSICC52343.2021.9420620
Armin Shoughi, M. B. Dowlatshahi
ECG beats have a key role in the reduction of fatality rate arising from cardiovascular diseases (CVDs) by using Arrhythmia diagnosis computer-aided systems and get the important information from patient cardiac conditions to the specialist. However, the accuracy and speed of arrhythmia diagnosis are challenging in ECG classification systems, and the existence of noise, instability nature, and imbalance in heartbeats challenged these systems. Accurate and on-time diagnosis of CVDs is a vital and important factor. So it has a significant effect on the treatment and recovery of patients. In this study, with the aim of accurate diagnosis of CVDs types, according to arrhythmia in ECG heartbeats, we implement an automatic ECG heartbeats classification by using discrete wavelet transformation on db2 mother wavelet and SMOTE oversampling algorithm as pre-processing level, and a classifier that consists of Convolutional neural network and BLSTM network. Then evaluate the proposed system on MIT-BIH imbalanced dataset, according to AAMI standards. The evaluations results show this approach with 50 epoch training achieved 99.78% accuracy for category F, 98.85% accuracy for category N, 99.43% accuracy for category S, 99.49% accuracy for category V, 99.87% accuracy for category Q. The source code is available at https://gitlab.com/arminshoughi/cnnlstmecg-classification. Our proposed classification system can be used as a tool for the automatic diagnosis of arrhythmia for CVDs specialists with the aim of primary screening of patients with heart arrhythmia.
{"title":"A practical system based on CNN-BLSTM network for accurate classification of ECG heartbeats of MIT-BIH imbalanced dataset","authors":"Armin Shoughi, M. B. Dowlatshahi","doi":"10.1109/CSICC52343.2021.9420620","DOIUrl":"https://doi.org/10.1109/CSICC52343.2021.9420620","url":null,"abstract":"ECG beats have a key role in the reduction of fatality rate arising from cardiovascular diseases (CVDs) by using Arrhythmia diagnosis computer-aided systems and get the important information from patient cardiac conditions to the specialist. However, the accuracy and speed of arrhythmia diagnosis are challenging in ECG classification systems, and the existence of noise, instability nature, and imbalance in heartbeats challenged these systems. Accurate and on-time diagnosis of CVDs is a vital and important factor. So it has a significant effect on the treatment and recovery of patients. In this study, with the aim of accurate diagnosis of CVDs types, according to arrhythmia in ECG heartbeats, we implement an automatic ECG heartbeats classification by using discrete wavelet transformation on db2 mother wavelet and SMOTE oversampling algorithm as pre-processing level, and a classifier that consists of Convolutional neural network and BLSTM network. Then evaluate the proposed system on MIT-BIH imbalanced dataset, according to AAMI standards. The evaluations results show this approach with 50 epoch training achieved 99.78% accuracy for category F, 98.85% accuracy for category N, 99.43% accuracy for category S, 99.49% accuracy for category V, 99.87% accuracy for category Q. The source code is available at https://gitlab.com/arminshoughi/cnnlstmecg-classification. Our proposed classification system can be used as a tool for the automatic diagnosis of arrhythmia for CVDs specialists with the aim of primary screening of patients with heart arrhythmia.","PeriodicalId":374593,"journal":{"name":"2021 26th International Computer Conference, Computer Society of Iran (CSICC)","volume":"54 23","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131875044","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 : 2021-03-03DOI: 10.1109/CSICC52343.2021.9420582
Narjes Farzi
today organizations encounter many issues such as newfound technologies, new business models, and rapid changes. That is, following the evolutions in the global context, caused by information and communication technology in the field of trade, industry, and specifically information technology, organizations, companies, and particularly banks have undergone changes and altered their reaction method to the market. In this way, the role of enterprise architecture and using standards and reference models are crucial to the organizations. Accordingly, organizations which want to be active in the digital transformation and move towards digital banking should be able to implement an agile enterprise architecture and use reference models such as BIAN. The objective of this article is to investigate the role of BIAN standard in moving towards digital banking.
{"title":"Investigation of the Place of BIAN Standard in Digital Banking Enterprise Architecture","authors":"Narjes Farzi","doi":"10.1109/CSICC52343.2021.9420582","DOIUrl":"https://doi.org/10.1109/CSICC52343.2021.9420582","url":null,"abstract":"today organizations encounter many issues such as newfound technologies, new business models, and rapid changes. That is, following the evolutions in the global context, caused by information and communication technology in the field of trade, industry, and specifically information technology, organizations, companies, and particularly banks have undergone changes and altered their reaction method to the market. In this way, the role of enterprise architecture and using standards and reference models are crucial to the organizations. Accordingly, organizations which want to be active in the digital transformation and move towards digital banking should be able to implement an agile enterprise architecture and use reference models such as BIAN. The objective of this article is to investigate the role of BIAN standard in moving towards digital banking.","PeriodicalId":374593,"journal":{"name":"2021 26th International Computer Conference, Computer Society of Iran (CSICC)","volume":"356 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116793085","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 : 2021-03-03DOI: 10.1109/CSICC52343.2021.9420548
Morteza Zakeri Nasrabadi, S. Parsa
Software testability is the propensity of code to reveal its existing faults, particularly during automated testing. Testing success depends on the testability of the program under test. On the other hand, testing success relies on the coverage of the test data provided by a given test data generation algorithm. However, little empirical evidence has been shown to clarify whether and how software testability affects test coverage. In this article, we propose a method to shed light on this subject. Our proposed framework uses the coverage of Software Under Test (SUT), provided by different automatically generated test suites, to build machine learning models, determining the testability of programs based on many source code metrics. The resultant models can predict the code coverage provided by a given test data generation algorithm before running the algorithm, reducing the cost of additional testing. The predicted coverage is used as a concrete proxy to quantify source code testability. Experiments show an acceptable accuracy of 81.94% in measuring and predicting software testability.
{"title":"Learning to Predict Software Testability","authors":"Morteza Zakeri Nasrabadi, S. Parsa","doi":"10.1109/CSICC52343.2021.9420548","DOIUrl":"https://doi.org/10.1109/CSICC52343.2021.9420548","url":null,"abstract":"Software testability is the propensity of code to reveal its existing faults, particularly during automated testing. Testing success depends on the testability of the program under test. On the other hand, testing success relies on the coverage of the test data provided by a given test data generation algorithm. However, little empirical evidence has been shown to clarify whether and how software testability affects test coverage. In this article, we propose a method to shed light on this subject. Our proposed framework uses the coverage of Software Under Test (SUT), provided by different automatically generated test suites, to build machine learning models, determining the testability of programs based on many source code metrics. The resultant models can predict the code coverage provided by a given test data generation algorithm before running the algorithm, reducing the cost of additional testing. The predicted coverage is used as a concrete proxy to quantify source code testability. Experiments show an acceptable accuracy of 81.94% in measuring and predicting software testability.","PeriodicalId":374593,"journal":{"name":"2021 26th International Computer Conference, Computer Society of Iran (CSICC)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130809461","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 : 2021-03-03DOI: 10.1109/CSICC52343.2021.9420558
Ahmad Jahanbin, M. S. Haghighi
Bitcoin, as the first and the most adopted cryptocurrency, offers many features one of which is contingent payment, that is, the owner of money can programmatically describe the condition upon which his/her money is spent. The condition is determined using a set of instructions written in the Bitcoin scripting language. Unfortunately, this scripting language is not sophisticated enough to create complex conditions or smart contracts in general. Many admirable efforts have been made to build a smart contract infrastructure on top of the Bitcoin platform. In this paper, given the inherent limitations of the Bitcoin scripting language, we critically analyze the practical effectiveness of these methods. Afterwards, we formally define what a smart contract is and introduce seven requirements that if are satisfied, can make creation of smart contracts for Bitcoin possible. Based on the introduced requirements, we examine the ability of the current methods that use secure Multi-party Computation (MPC) to create smart contracts for Bitcoin and show where they fall short. We additionally compare their pros and cons and give clues on how a comprehensive smart contract platform can be possibly built for Bitcoin.
{"title":"On the Possibility of Creating Smart Contracts on Bitcoin by MPC-based Approaches","authors":"Ahmad Jahanbin, M. S. Haghighi","doi":"10.1109/CSICC52343.2021.9420558","DOIUrl":"https://doi.org/10.1109/CSICC52343.2021.9420558","url":null,"abstract":"Bitcoin, as the first and the most adopted cryptocurrency, offers many features one of which is contingent payment, that is, the owner of money can programmatically describe the condition upon which his/her money is spent. The condition is determined using a set of instructions written in the Bitcoin scripting language. Unfortunately, this scripting language is not sophisticated enough to create complex conditions or smart contracts in general. Many admirable efforts have been made to build a smart contract infrastructure on top of the Bitcoin platform. In this paper, given the inherent limitations of the Bitcoin scripting language, we critically analyze the practical effectiveness of these methods. Afterwards, we formally define what a smart contract is and introduce seven requirements that if are satisfied, can make creation of smart contracts for Bitcoin possible. Based on the introduced requirements, we examine the ability of the current methods that use secure Multi-party Computation (MPC) to create smart contracts for Bitcoin and show where they fall short. We additionally compare their pros and cons and give clues on how a comprehensive smart contract platform can be possibly built for Bitcoin.","PeriodicalId":374593,"journal":{"name":"2021 26th International Computer Conference, Computer Society of Iran (CSICC)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128718645","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 : 2021-03-03DOI: 10.1109/CSICC52343.2021.9420630
Ali Alemi Matin Pour, S. Jalili
Extracting aspect term is essential for aspect level sentiment analysis; Sentiment analysis collects and extracts the opinions expressed in social media and websites' comments and then analyzes them, helping users and stakeholders understand public views on the issues raised better and more quickly. Aspect-level sentiment analysis provides more detailed information, which is very beneficial for use in many various domains. In this paper, the significant contribution is to provide a data preprocessing method and a deep convolutional neural network (CNN) to label each word in opinionated sentences as an aspect or non-aspect word. The proposed method extracts the terms of the aspect that can be used in analyzing the sentiment of the expressed aspect terms in the comments and opinions. The experimental results of the proposed method performed on the SemEval-2014 dataset show that it performs better than other prominent methods such as deep CNN. The proposed data preprocessing method with the deep CNN network can improve extraction of aspect terms according to F-measure by at least 1.05% and 0.95% on restaurant and laptop domains.
{"title":"Aspects Extraction for Aspect Level Opinion Analysis Based on Deep CNN","authors":"Ali Alemi Matin Pour, S. Jalili","doi":"10.1109/CSICC52343.2021.9420630","DOIUrl":"https://doi.org/10.1109/CSICC52343.2021.9420630","url":null,"abstract":"Extracting aspect term is essential for aspect level sentiment analysis; Sentiment analysis collects and extracts the opinions expressed in social media and websites' comments and then analyzes them, helping users and stakeholders understand public views on the issues raised better and more quickly. Aspect-level sentiment analysis provides more detailed information, which is very beneficial for use in many various domains. In this paper, the significant contribution is to provide a data preprocessing method and a deep convolutional neural network (CNN) to label each word in opinionated sentences as an aspect or non-aspect word. The proposed method extracts the terms of the aspect that can be used in analyzing the sentiment of the expressed aspect terms in the comments and opinions. The experimental results of the proposed method performed on the SemEval-2014 dataset show that it performs better than other prominent methods such as deep CNN. The proposed data preprocessing method with the deep CNN network can improve extraction of aspect terms according to F-measure by at least 1.05% and 0.95% on restaurant and laptop domains.","PeriodicalId":374593,"journal":{"name":"2021 26th International Computer Conference, Computer Society of Iran (CSICC)","volume":"296 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127179984","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：