Pub Date : 2022-01-26DOI: 10.1109/KST53302.2022.9729075
Khin Me Me Chit, Yi Yi Chan Myae Win Shein, Wai Yan, A. Khine
Hate Speech on Social Media is definitely an evolving threat for every nation, especially for countries like Myanmar. Lack of media and digital literacy is playing a huge role in making people insult to each other or misallocating their stresses to others without physical encounter. Moreover, disingenuous politicians fuel online hate speech campaigns backstage of the elections by targeting different religions in the regard of heretics and using racialism. To emphasize this matter, we scraped over 16,000 social media comments from the most popular social media platform in Myanmar and performed hate-speech research using those samples. With the precise definition of a hate speech labelling guideline, annotation on the sample dataset was done systematically and efficiently. Experiments and evaluations were conducted using different linear and non-linear deep-learning classification models. Performances of the models are at the peak in Logistic Regression among linear models with 0.8974 AUC score and XLM-RoBERTa among deep learning models with 0.8958 AUC score on the test dataset. We observed that it is more advantageous to use linear models on our dataset since they achieved comparable results to the deep learning models and have much lower computational cost.
{"title":"SIREN! Detecting Burmese Hate Speech Comments on Social Media","authors":"Khin Me Me Chit, Yi Yi Chan Myae Win Shein, Wai Yan, A. Khine","doi":"10.1109/KST53302.2022.9729075","DOIUrl":"https://doi.org/10.1109/KST53302.2022.9729075","url":null,"abstract":"Hate Speech on Social Media is definitely an evolving threat for every nation, especially for countries like Myanmar. Lack of media and digital literacy is playing a huge role in making people insult to each other or misallocating their stresses to others without physical encounter. Moreover, disingenuous politicians fuel online hate speech campaigns backstage of the elections by targeting different religions in the regard of heretics and using racialism. To emphasize this matter, we scraped over 16,000 social media comments from the most popular social media platform in Myanmar and performed hate-speech research using those samples. With the precise definition of a hate speech labelling guideline, annotation on the sample dataset was done systematically and efficiently. Experiments and evaluations were conducted using different linear and non-linear deep-learning classification models. Performances of the models are at the peak in Logistic Regression among linear models with 0.8974 AUC score and XLM-RoBERTa among deep learning models with 0.8958 AUC score on the test dataset. We observed that it is more advantageous to use linear models on our dataset since they achieved comparable results to the deep learning models and have much lower computational cost.","PeriodicalId":433638,"journal":{"name":"2022 14th International Conference on Knowledge and Smart Technology (KST)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123120903","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 : 2022-01-26DOI: 10.1109/KST53302.2022.9727233
Tantep Sinjanakhom, S. Chivapreecha
This This paper presents a real-time signal processing system in which a neural network generates the impulse response (IR) of a Marshall 1960A guitar cabinet with 25W Celestion speakers based on user-specified parameters. The parameters include the microphone type, position of the speaker on which the microphone is mounted, distance between the microphone and the cabinet, and off-axis tilting angle. The trained model of neural network can generate the impulse response for a speaker cabinet, as well as the sound of settings not included in training set. Cross-correlation, error-to-signal ratio, power spectral density error, and magnitude-squared coherence were all utilized to assess the model's output. Mean Opinion Score listening tests were performed to determine the similarity of the convolved guitar signals. According to the results, the emulated cabinet sounds were perceived to be nearly identical to the original sounds. The performance of the real-time audio plugin implementation is proved to be computationally efficient. Because raw IR data for each microphone configuration does not need to be saved directly to the PC's memory, utilizing it in music production work can be more convenient, allowing the user to modify the parameters while hearing the differences without having to repeat the IR file loading procedure.
{"title":"Neural Networks for Real-Time Digital Emulation of Guitar Speaker Cabinet Impulse Response","authors":"Tantep Sinjanakhom, S. Chivapreecha","doi":"10.1109/KST53302.2022.9727233","DOIUrl":"https://doi.org/10.1109/KST53302.2022.9727233","url":null,"abstract":"This This paper presents a real-time signal processing system in which a neural network generates the impulse response (IR) of a Marshall 1960A guitar cabinet with 25W Celestion speakers based on user-specified parameters. The parameters include the microphone type, position of the speaker on which the microphone is mounted, distance between the microphone and the cabinet, and off-axis tilting angle. The trained model of neural network can generate the impulse response for a speaker cabinet, as well as the sound of settings not included in training set. Cross-correlation, error-to-signal ratio, power spectral density error, and magnitude-squared coherence were all utilized to assess the model's output. Mean Opinion Score listening tests were performed to determine the similarity of the convolved guitar signals. According to the results, the emulated cabinet sounds were perceived to be nearly identical to the original sounds. The performance of the real-time audio plugin implementation is proved to be computationally efficient. Because raw IR data for each microphone configuration does not need to be saved directly to the PC's memory, utilizing it in music production work can be more convenient, allowing the user to modify the parameters while hearing the differences without having to repeat the IR file loading procedure.","PeriodicalId":433638,"journal":{"name":"2022 14th International Conference on Knowledge and Smart Technology (KST)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126906713","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 : 2022-01-26DOI: 10.1109/KST53302.2022.9729071
Wandee Aunsa-Ard, T. Kerdcharoen
The coffee industry is facing increasing challenges due to climate change, pests, diseases, which leads to the reduced production and negative impact on coffee qualities. Thus, quality assurance of coffee from production to roasting and brewing becomes more important, especially coffee flavor and aroma. This research aims to study the applicability of electronic nose (e-nose) and algorithm to detect coffee aroma obtained from different origins. The coffee beans used in this experiment were obtained from different areas in northern Thailand. These coffee beans have different growing conditions, altitude, processing and roasting condition. In this study, the three aspects of e-nose were investigated; (i) e-nose sensitivity to coffee odors, (ii) e-nose capability of correctly recognizing the detected odors and (iii) factors that influence coffee odors such as altitude, processing and roasting conditions. The e-nose system comprises of eight metal oxide semiconductor (MOX) gas sensors and in-house developed analysis software. Principal Component Analysis (PCA) is a classification algorithm for pattern recognition of different coffee aroma. Based on experimental results, the e-nose technology shows a capability to detect and distinguish the coffee odors caused by different altitude, processing and roasting process. E-nose is a suitable method for aroma detection in coffee industry to enhance the quality.
{"title":"Electronic Nose for Analysis of Coffee Beans Obtained from Different Altitudes and Origin","authors":"Wandee Aunsa-Ard, T. Kerdcharoen","doi":"10.1109/KST53302.2022.9729071","DOIUrl":"https://doi.org/10.1109/KST53302.2022.9729071","url":null,"abstract":"The coffee industry is facing increasing challenges due to climate change, pests, diseases, which leads to the reduced production and negative impact on coffee qualities. Thus, quality assurance of coffee from production to roasting and brewing becomes more important, especially coffee flavor and aroma. This research aims to study the applicability of electronic nose (e-nose) and algorithm to detect coffee aroma obtained from different origins. The coffee beans used in this experiment were obtained from different areas in northern Thailand. These coffee beans have different growing conditions, altitude, processing and roasting condition. In this study, the three aspects of e-nose were investigated; (i) e-nose sensitivity to coffee odors, (ii) e-nose capability of correctly recognizing the detected odors and (iii) factors that influence coffee odors such as altitude, processing and roasting conditions. The e-nose system comprises of eight metal oxide semiconductor (MOX) gas sensors and in-house developed analysis software. Principal Component Analysis (PCA) is a classification algorithm for pattern recognition of different coffee aroma. Based on experimental results, the e-nose technology shows a capability to detect and distinguish the coffee odors caused by different altitude, processing and roasting process. E-nose is a suitable method for aroma detection in coffee industry to enhance the quality.","PeriodicalId":433638,"journal":{"name":"2022 14th International Conference on Knowledge and Smart Technology (KST)","volume":"94 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122026471","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 : 2022-01-26DOI: 10.1109/KST53302.2022.9727234
Huey-Ing Liu, Meng-Wei Chen, Wei-Chun Kao, Yao-Wen Yeh, Cheng Yang
This paper proposes a hybrid Gated Recurrent Unit (GRU) and Self-Attention based model, named GSAP, for dual medical related NLP tasks. GSAP stacks three famous neural network units: GRU, self-attention and pooling of Convolutional Neural Network (CNN) to improve the accuracy. In the GSAP, GRU is first adopted to comprehend sentences. Second, the Self-Attention layer helps the model to focus on key points of inputs. Finally, the pooling layer eases the outfitting and upgrades the system accuracy. The proposed GSAP is applied to two different medical NLP tasks: medical QA matching and smoking status classification and demonstrates outstanding results. In the smoking prediction, GSAP obtains an accuracy around 80%. Regarding to the medical QA matching task, GSAP upgrades the accuracy up to around 90%.
{"title":"GSAP: A Hybrid GRU and Self-Attention Based Model for Dual Medical NLP Tasks","authors":"Huey-Ing Liu, Meng-Wei Chen, Wei-Chun Kao, Yao-Wen Yeh, Cheng Yang","doi":"10.1109/KST53302.2022.9727234","DOIUrl":"https://doi.org/10.1109/KST53302.2022.9727234","url":null,"abstract":"This paper proposes a hybrid Gated Recurrent Unit (GRU) and Self-Attention based model, named GSAP, for dual medical related NLP tasks. GSAP stacks three famous neural network units: GRU, self-attention and pooling of Convolutional Neural Network (CNN) to improve the accuracy. In the GSAP, GRU is first adopted to comprehend sentences. Second, the Self-Attention layer helps the model to focus on key points of inputs. Finally, the pooling layer eases the outfitting and upgrades the system accuracy. The proposed GSAP is applied to two different medical NLP tasks: medical QA matching and smoking status classification and demonstrates outstanding results. In the smoking prediction, GSAP obtains an accuracy around 80%. Regarding to the medical QA matching task, GSAP upgrades the accuracy up to around 90%.","PeriodicalId":433638,"journal":{"name":"2022 14th International Conference on Knowledge and Smart Technology (KST)","volume":"223 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116164814","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 : 2022-01-26DOI: 10.1109/KST53302.2022.9729068
Hahyeon Kim, Chen Li
In this paper, we present the concept of a robot service that interacts with a human operator through speech. Service-oriented architectures and cloud computing are now the dominant computer paradigms. Research on using robots as a service (Robot-as-a-Service. RaaS) is a new trend based on the integration of robots and embedded devices and web and cloud computing. However, while the demand for RaaS utilization within the industry is high, it has not successfully acquired many users. One of the reasons is that the accuracy of the robot's recognition-judgment-action process does not reach a level that users can trust, and the other reason is that it is challenging to learn how to control the robot. Therefore, this study focused on services that allow users to control robots easily within industrial sites. Speech recognition was implemented using RESTful API and Server-Client communication, and a mobile manipulator robot, what we call Little-Helper (LH), is used for implementation. According to the human operators, communicate with the robot by voice speech, increasing collaboration efficiency and productivity of the industry is expected.
{"title":"RaaS (Robot-as-a-Service) focusing on the human-robot collaboration in industrial sites","authors":"Hahyeon Kim, Chen Li","doi":"10.1109/KST53302.2022.9729068","DOIUrl":"https://doi.org/10.1109/KST53302.2022.9729068","url":null,"abstract":"In this paper, we present the concept of a robot service that interacts with a human operator through speech. Service-oriented architectures and cloud computing are now the dominant computer paradigms. Research on using robots as a service (Robot-as-a-Service. RaaS) is a new trend based on the integration of robots and embedded devices and web and cloud computing. However, while the demand for RaaS utilization within the industry is high, it has not successfully acquired many users. One of the reasons is that the accuracy of the robot's recognition-judgment-action process does not reach a level that users can trust, and the other reason is that it is challenging to learn how to control the robot. Therefore, this study focused on services that allow users to control robots easily within industrial sites. Speech recognition was implemented using RESTful API and Server-Client communication, and a mobile manipulator robot, what we call Little-Helper (LH), is used for implementation. According to the human operators, communicate with the robot by voice speech, increasing collaboration efficiency and productivity of the industry is expected.","PeriodicalId":433638,"journal":{"name":"2022 14th International Conference on Knowledge and Smart Technology (KST)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134006789","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 : 2022-01-26DOI: 10.1109/KST53302.2022.9729086
Thaninthorn Whasphutthisit, Watchareewan Jitsakul
This paper presents to compare prediction models for road deaths on road network by data mining techniques. In this work, the classifier is selected from four prediction algorithms: Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Neural Network (NN). The dead injured and dead people data in road accident data set of the Ministry of Transport, Thailand from January to April 2021. It has up to 8,560 records 46 attributes. This research has measured performance models with accuracy, precision, recall, and f-measure. The comparative results showed that the accuracy of RF is the most appropriate for predicting road deaths on road network with accuracy 89%, precision 0.86, recall 0.89, and f-measure 0.85.
本文采用数据挖掘技术对路网道路死亡预测模型进行了比较。在这项工作中，分类器从四种预测算法中选择:随机森林(RF)，支持向量机(SVM)， k -最近邻(KNN)和神经网络(NN)。泰国交通部2021年1 - 4月道路交通事故数据集中的死伤人数数据。它有多达8,560条记录46个属性。本研究用准确性、精密度、召回率和f-measure来衡量绩效模型。对比结果表明，RF预测路网道路死亡的准确度为89%，精密度为0.86，召回率为0.89,f-measure为0.85。
{"title":"Comparison of Prediction Models for Road Deaths On Road Network","authors":"Thaninthorn Whasphutthisit, Watchareewan Jitsakul","doi":"10.1109/KST53302.2022.9729086","DOIUrl":"https://doi.org/10.1109/KST53302.2022.9729086","url":null,"abstract":"This paper presents to compare prediction models for road deaths on road network by data mining techniques. In this work, the classifier is selected from four prediction algorithms: Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Neural Network (NN). The dead injured and dead people data in road accident data set of the Ministry of Transport, Thailand from January to April 2021. It has up to 8,560 records 46 attributes. This research has measured performance models with accuracy, precision, recall, and f-measure. The comparative results showed that the accuracy of RF is the most appropriate for predicting road deaths on road network with accuracy 89%, precision 0.86, recall 0.89, and f-measure 0.85.","PeriodicalId":433638,"journal":{"name":"2022 14th International Conference on Knowledge and Smart Technology (KST)","volume":"428 12","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133941274","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 : 2022-01-26DOI: 10.1109/KST53302.2022.9729061
Donghun Yang, Myunggwon Hwang
Bigdata and IoT technologies are developing rapidly. Accordingly, consideration of network security is also emphasized, and efficient intrusion detection technology is required for detecting increasingly sophisticated network attacks. In this study, we propose an efficient network anomaly detection method based on ensemble and unsupervised learning. The proposed model is built by training an autoencoder, a representative unsupervised deep learning model, using only normal network traffic data. The anomaly score of the detection target data is derived by ensemble the reconstruction loss and the Mahalanobis distances for each layer output of the trained autoencoder. By applying a threshold to this score, network anomaly traffic can be efficiently detected. To evaluate the proposed model, we applied our method to UNSW-NB15 dataset. The results show that the overall performance of the proposed method is superior to those of the model using only the reconstruction loss of the autoencoder and the model applying the Mahalanobis distance to the raw data.
{"title":"Unsupervised and Ensemble-based Anomaly Detection Method for Network Security","authors":"Donghun Yang, Myunggwon Hwang","doi":"10.1109/KST53302.2022.9729061","DOIUrl":"https://doi.org/10.1109/KST53302.2022.9729061","url":null,"abstract":"Bigdata and IoT technologies are developing rapidly. Accordingly, consideration of network security is also emphasized, and efficient intrusion detection technology is required for detecting increasingly sophisticated network attacks. In this study, we propose an efficient network anomaly detection method based on ensemble and unsupervised learning. The proposed model is built by training an autoencoder, a representative unsupervised deep learning model, using only normal network traffic data. The anomaly score of the detection target data is derived by ensemble the reconstruction loss and the Mahalanobis distances for each layer output of the trained autoencoder. By applying a threshold to this score, network anomaly traffic can be efficiently detected. To evaluate the proposed model, we applied our method to UNSW-NB15 dataset. The results show that the overall performance of the proposed method is superior to those of the model using only the reconstruction loss of the autoencoder and the model applying the Mahalanobis distance to the raw data.","PeriodicalId":433638,"journal":{"name":"2022 14th International Conference on Knowledge and Smart Technology (KST)","volume":"127 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114276027","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 : 2022-01-26DOI: 10.1109/KST53302.2022.9727231
Jun-Hao Chen, Cheng-Han Wu, Yun-Cheng Tsai, Samuel Yen-Chi Chen
More and more hedge funds have integrated AI techniques into the traditional trading strategy to speculate on Digital Currency. Among the conventional technical analysis, candlestick pattern recognition is a critical financial trading technique by visual judgment on graphical price movement. A model with high accuracy still can not meet the demand under the highly regulated financial industry that requires understanding the decision-making and quantifying the potential risk. Despite the deep convolutional neural networks (CNNs) have a significant performance. Especially in a highly speculative market, blindly trusting a black-box model will incur lots of troubles. Therefore, it is necessary to incorporate explainability into a DNN-based classic trading strategy, candlestick pattern recognition. It can make an acceptable justification for traders in the Digital Currency market. The paper exposes the black box and provides two algorithms as following. The first is an Adversarial Interpreter to explore the explainability. The second is an Adversarial Generator to enhance the model's explainability. To trust in the AI model and understand its judgment, the participant adopts powerful AI techniques to create more possibilities for AI in the Digital Currency market.
{"title":"Explainable Digital Currency Candlestick Pattern AI Learner","authors":"Jun-Hao Chen, Cheng-Han Wu, Yun-Cheng Tsai, Samuel Yen-Chi Chen","doi":"10.1109/KST53302.2022.9727231","DOIUrl":"https://doi.org/10.1109/KST53302.2022.9727231","url":null,"abstract":"More and more hedge funds have integrated AI techniques into the traditional trading strategy to speculate on Digital Currency. Among the conventional technical analysis, candlestick pattern recognition is a critical financial trading technique by visual judgment on graphical price movement. A model with high accuracy still can not meet the demand under the highly regulated financial industry that requires understanding the decision-making and quantifying the potential risk. Despite the deep convolutional neural networks (CNNs) have a significant performance. Especially in a highly speculative market, blindly trusting a black-box model will incur lots of troubles. Therefore, it is necessary to incorporate explainability into a DNN-based classic trading strategy, candlestick pattern recognition. It can make an acceptable justification for traders in the Digital Currency market. The paper exposes the black box and provides two algorithms as following. The first is an Adversarial Interpreter to explore the explainability. The second is an Adversarial Generator to enhance the model's explainability. To trust in the AI model and understand its judgment, the participant adopts powerful AI techniques to create more possibilities for AI in the Digital Currency market.","PeriodicalId":433638,"journal":{"name":"2022 14th International Conference on Knowledge and Smart Technology (KST)","volume":"73 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122607339","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 : 2022-01-26DOI: 10.1109/KST53302.2022.9729056
Pat Vatiwutipong
The Radon transform is one of the most well-known tools for line detection methods. The drawback of the Radon transform is that it is not continuous. So slight change in an image may lead to a considerable difference in the detection line. This is undesirable. We solve this problem by the modified version of the Radon transform called the d-Radon transform. Several mathematical properties of this modified transform, especially a continuity property of line detection methods, were studied. We focus on a space of image containing several sizes of circles. A metric function on that space is proposed to measure the change of images. By this new transformation, continuity property is obtained.
{"title":"Continuity of line detection methods based on the Radon transform","authors":"Pat Vatiwutipong","doi":"10.1109/KST53302.2022.9729056","DOIUrl":"https://doi.org/10.1109/KST53302.2022.9729056","url":null,"abstract":"The Radon transform is one of the most well-known tools for line detection methods. The drawback of the Radon transform is that it is not continuous. So slight change in an image may lead to a considerable difference in the detection line. This is undesirable. We solve this problem by the modified version of the Radon transform called the d-Radon transform. Several mathematical properties of this modified transform, especially a continuity property of line detection methods, were studied. We focus on a space of image containing several sizes of circles. A metric function on that space is proposed to measure the change of images. By this new transformation, continuity property is obtained.","PeriodicalId":433638,"journal":{"name":"2022 14th International Conference on Knowledge and Smart Technology (KST)","volume":"180 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123008239","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 : 2022-01-26DOI: 10.1109/KST53302.2022.9729060
Watcharachat Plangsri, Nalina Phisanbut, P. Piamsa-nga
Research documents play a crucial role in data-driven research. Identifying concepts in a corpus of research documents can lead to a better understanding of the current stage of research. It can reveal fruitful concepts hidden inside the corpus. However, manually analyzing the corpus is laborious and inefficient. Automating the process is challenging due to the lack of background knowledge to fill the semantic gap that exists between humans and machines. To address this issue, we introduce a novel method that leverages information from an online open resource, namely Wikipedia, to build background knowledge automatically. An experiment on a set of 13,636 research documents shows that the framework can effectively and efficiently identify broad range of concepts within a large text corpus by exploiting only Wikipedia categories and documents' titles.
{"title":"Unsupervised concept identification from a large corpus of research documents","authors":"Watcharachat Plangsri, Nalina Phisanbut, P. Piamsa-nga","doi":"10.1109/KST53302.2022.9729060","DOIUrl":"https://doi.org/10.1109/KST53302.2022.9729060","url":null,"abstract":"Research documents play a crucial role in data-driven research. Identifying concepts in a corpus of research documents can lead to a better understanding of the current stage of research. It can reveal fruitful concepts hidden inside the corpus. However, manually analyzing the corpus is laborious and inefficient. Automating the process is challenging due to the lack of background knowledge to fill the semantic gap that exists between humans and machines. To address this issue, we introduce a novel method that leverages information from an online open resource, namely Wikipedia, to build background knowledge automatically. An experiment on a set of 13,636 research documents shows that the framework can effectively and efficiently identify broad range of concepts within a large text corpus by exploiting only Wikipedia categories and documents' titles.","PeriodicalId":433638,"journal":{"name":"2022 14th International Conference on Knowledge and Smart Technology (KST)","volume":"46 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114685097","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：