Pub Date : 2020-12-22DOI: 10.1109/IKT51791.2020.9345617
Tahereh Zarrat Ehsan, S. M. Mohtavipour
Video surveillance cameras are widely used due to security concerns. Analyzing these large amounts of videos by a human operator is a difficult and time-consuming job. To overcome this problem, automatic violence detection in video sequences has become an active research area of computer vision in recent years. Early methods focused on hand-engineering approaches to construct hand-crafted features, but they are not discriminative enough for complex actions like violence. To extract complex behavioral features automatically, it is required to apply deep networks. In this paper, we proposed a novel Vi-Net architecture based on the deep Convolutional Neural Network (CNN) to detect actions with abnormal velocity. Motion patterns of targets in the video are estimated by optical flow vectors to train the Vi-Net network. As violent behavior comprises fast movements, these vectors are useful for the extraction of distinctive features. We performed several experiments on Hockey, Crowd, and Movies datasets and results showed that the proposed architecture achieved higher accuracy in comparison with the state-of-the-art methods.
{"title":"Vi-Net: A Deep Violent Flow Network for Violence Detection in Video Sequences","authors":"Tahereh Zarrat Ehsan, S. M. Mohtavipour","doi":"10.1109/IKT51791.2020.9345617","DOIUrl":"https://doi.org/10.1109/IKT51791.2020.9345617","url":null,"abstract":"Video surveillance cameras are widely used due to security concerns. Analyzing these large amounts of videos by a human operator is a difficult and time-consuming job. To overcome this problem, automatic violence detection in video sequences has become an active research area of computer vision in recent years. Early methods focused on hand-engineering approaches to construct hand-crafted features, but they are not discriminative enough for complex actions like violence. To extract complex behavioral features automatically, it is required to apply deep networks. In this paper, we proposed a novel Vi-Net architecture based on the deep Convolutional Neural Network (CNN) to detect actions with abnormal velocity. Motion patterns of targets in the video are estimated by optical flow vectors to train the Vi-Net network. As violent behavior comprises fast movements, these vectors are useful for the extraction of distinctive features. We performed several experiments on Hockey, Crowd, and Movies datasets and results showed that the proposed architecture achieved higher accuracy in comparison with the state-of-the-art methods.","PeriodicalId":382725,"journal":{"name":"2020 11th International Conference on Information and Knowledge Technology (IKT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129799222","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 : 2020-12-22DOI: 10.1109/IKT51791.2020.9345628
Hadi Aghaee, Bahareh Akhbari
In this paper, the classical-quantum multiple access wiretap channel with a common message is studied under the one-shot setting. In this regard, an inner bound is derived using simultaneous decoding. One important problem in multi-terminal quantum networks is the nonexistence of a proven simultaneous decoder for decoding more than two messages simultaneously. The main focus of this paper is to construct a simultaneous decoder for the one-shot setting.
{"title":"Classical-Quantum Multiple Access Wiretap Channel with Common Message: One-Shot Rate Region","authors":"Hadi Aghaee, Bahareh Akhbari","doi":"10.1109/IKT51791.2020.9345628","DOIUrl":"https://doi.org/10.1109/IKT51791.2020.9345628","url":null,"abstract":"In this paper, the classical-quantum multiple access wiretap channel with a common message is studied under the one-shot setting. In this regard, an inner bound is derived using simultaneous decoding. One important problem in multi-terminal quantum networks is the nonexistence of a proven simultaneous decoder for decoding more than two messages simultaneously. The main focus of this paper is to construct a simultaneous decoder for the one-shot setting.","PeriodicalId":382725,"journal":{"name":"2020 11th International Conference on Information and Knowledge Technology (IKT)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114259727","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 : 2020-12-22DOI: 10.1109/IKT51791.2020.9345622
M. D. Khomami, Alireza Rezvanian, A. Saghiri, M. Meybodi
Maximal clique finding is a fundamental problem in graph theory and has been broadly investigated. However, maximal clique finding is time-consuming due to its nature and always returns tremendous cliques with large overlap nodes. Hence, a solution uses the relaxed version of the clique called k-clique, which follows up the subset of vertices with size k such that each pair in this subset has an edge. The k-clique problem has several applications in different domains, such as motif detection, finding anomalies in large graphs, and community structure discovery. In this paper, an algorithm based on learning automata is proposed for finding k-clique called (KC-LA) to apply communities in complex social networks. In (KC-LA), a network of learning automata is considering to the underlying networks. Then, select the proper action from a set of allowable actions, the reward and penalty guide KC-LA to detect the k-clique. Also, we applied the k-clique in the concept of finding communities in complex social networks. The KC-LA algorithm is to design some breakthroughs on the real and synthetic graphs in terms of high efficiency and effectiveness.
{"title":"Distributed Learning Automata-Based Algorithm for Finding K-Clique in Complex Social Networks","authors":"M. D. Khomami, Alireza Rezvanian, A. Saghiri, M. Meybodi","doi":"10.1109/IKT51791.2020.9345622","DOIUrl":"https://doi.org/10.1109/IKT51791.2020.9345622","url":null,"abstract":"Maximal clique finding is a fundamental problem in graph theory and has been broadly investigated. However, maximal clique finding is time-consuming due to its nature and always returns tremendous cliques with large overlap nodes. Hence, a solution uses the relaxed version of the clique called k-clique, which follows up the subset of vertices with size k such that each pair in this subset has an edge. The k-clique problem has several applications in different domains, such as motif detection, finding anomalies in large graphs, and community structure discovery. In this paper, an algorithm based on learning automata is proposed for finding k-clique called (KC-LA) to apply communities in complex social networks. In (KC-LA), a network of learning automata is considering to the underlying networks. Then, select the proper action from a set of allowable actions, the reward and penalty guide KC-LA to detect the k-clique. Also, we applied the k-clique in the concept of finding communities in complex social networks. The KC-LA algorithm is to design some breakthroughs on the real and synthetic graphs in terms of high efficiency and effectiveness.","PeriodicalId":382725,"journal":{"name":"2020 11th International Conference on Information and Knowledge Technology (IKT)","volume":"103 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116175422","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 : 2020-12-22DOI: 10.1109/IKT51791.2020.9345608
S. M. Mohtavipour, H. Shahhoseini
High-performance computing systems including Reconfigurable Hardware (RH) such as Field Programmable Gate Array (FPGA) proved a significant impact on the speed of application execution with useful reconfiguration and parallelism attributes. To make one application executable on RH, it is required to perform some heavy computational compilation preprocessing phases. In this paper, we aim to reduce compilation overhead in the NP-hard problem of the mapping phase by utilizing a novel Parallelized Genetic Algorithm (PGA) which is based on potential solutions in the search space. In the search space of possible solutions, we analytically separate weak and potential solutions to guide the GA for reaching the optimal solution faster. Moreover, this separation has been carried out independently to add parallelism into our GA and also, to switch between search spaces for keeping the generalization of GA exploration. Comparison results showed that our approach could make a considerable gap at the starting points of solution searching and therefore, found the optimal solution in a more reasonable time.
{"title":"A Potential Solutions-Based Parallelized GA for Application Graph Mapping in Reconfigurable Hardware","authors":"S. M. Mohtavipour, H. Shahhoseini","doi":"10.1109/IKT51791.2020.9345608","DOIUrl":"https://doi.org/10.1109/IKT51791.2020.9345608","url":null,"abstract":"High-performance computing systems including Reconfigurable Hardware (RH) such as Field Programmable Gate Array (FPGA) proved a significant impact on the speed of application execution with useful reconfiguration and parallelism attributes. To make one application executable on RH, it is required to perform some heavy computational compilation preprocessing phases. In this paper, we aim to reduce compilation overhead in the NP-hard problem of the mapping phase by utilizing a novel Parallelized Genetic Algorithm (PGA) which is based on potential solutions in the search space. In the search space of possible solutions, we analytically separate weak and potential solutions to guide the GA for reaching the optimal solution faster. Moreover, this separation has been carried out independently to add parallelism into our GA and also, to switch between search spaces for keeping the generalization of GA exploration. Comparison results showed that our approach could make a considerable gap at the starting points of solution searching and therefore, found the optimal solution in a more reasonable time.","PeriodicalId":382725,"journal":{"name":"2020 11th International Conference on Information and Knowledge Technology (IKT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128691815","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 : 2020-12-22DOI: 10.1109/ikt51791.2020.9345624
Golshan Assadat Afzali Boroujeni, Heshaam Faili
Answer selection in community question answering is a challenging task in natural language processing. The main problem is that there is no evaluation for the answers given by the users and one should go through all possible answers for assessing them, which is exhausting and time consuming. In this paper $wtext{e}$ propose a latent-variable model for learning the representations of the question and answer, by jointly optimizing generative and discriminative objectives. This model uses variational autoencoders (VAE) in a multi-task learning process with a classifier to produces a representation for each answer by which the classifier could classify it's relation with correspond question with a high performance. The experimental results on two public datasets, SemEval 2015 and SemEval 2017, recognize the significance of the proposed framework, especially for the semi-supervised setting. The results showed that the proposed model outperformed F1 of state-of-the-art method up to about 8% for SemEval 2015 and about 5% for SemEva1 2017.
{"title":"Using Deconvolutional Variational Autoencoder for Answer Selection in Community Question Answering","authors":"Golshan Assadat Afzali Boroujeni, Heshaam Faili","doi":"10.1109/ikt51791.2020.9345624","DOIUrl":"https://doi.org/10.1109/ikt51791.2020.9345624","url":null,"abstract":"Answer selection in community question answering is a challenging task in natural language processing. The main problem is that there is no evaluation for the answers given by the users and one should go through all possible answers for assessing them, which is exhausting and time consuming. In this paper $wtext{e}$ propose a latent-variable model for learning the representations of the question and answer, by jointly optimizing generative and discriminative objectives. This model uses variational autoencoders (VAE) in a multi-task learning process with a classifier to produces a representation for each answer by which the classifier could classify it's relation with correspond question with a high performance. The experimental results on two public datasets, SemEval 2015 and SemEval 2017, recognize the significance of the proposed framework, especially for the semi-supervised setting. The results showed that the proposed model outperformed F1 of state-of-the-art method up to about 8% for SemEval 2015 and about 5% for SemEva1 2017.","PeriodicalId":382725,"journal":{"name":"2020 11th International Conference on Information and Knowledge Technology (IKT)","volume":"31 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123521487","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 : 2020-12-22DOI: 10.1109/IKT51791.2020.9345625
Farkhondeh Kiaee
The vehicle-to-grid (V2G) technology provides an opportunity to generate revenue by selling electricity back to the grid at peak times when electricity is more expensive. Instead of sharing a contaminated pump handle at a gas station during the current covid-19 pandemic, plugging in the electric vehicle (EV) at home makes feel much safer. A V2G control algorithm is necessary to decide whether the electric vehicle (EV) should be charged or discharged in each hour. In this paper, we study the real-time V2G control problem under price uncertainty where the electricity price is determined dynamically every hour. Our model is inspired by the Deep Q-learning (DQN) algorithm which combines popular Q-learning with a deep neural network. The proposed Double-DQN model is an update of the DQN which maintains two distinct networks to select or evaluate an action. The Double-DQN algorithm is used to control charge/discharge operation in the hourly available electricity price in order to maximize the profit for the EV owner during the whole parking time. Experiment results show that our proposed method can work effectively in the real electricity market and it is able to increase the profit significantly compared with the other state-of-the-art EV charging schemes.
{"title":"Integration of Electric Vehicles in Smart Grid using Deep Reinforcement Learning","authors":"Farkhondeh Kiaee","doi":"10.1109/IKT51791.2020.9345625","DOIUrl":"https://doi.org/10.1109/IKT51791.2020.9345625","url":null,"abstract":"The vehicle-to-grid (V2G) technology provides an opportunity to generate revenue by selling electricity back to the grid at peak times when electricity is more expensive. Instead of sharing a contaminated pump handle at a gas station during the current covid-19 pandemic, plugging in the electric vehicle (EV) at home makes feel much safer. A V2G control algorithm is necessary to decide whether the electric vehicle (EV) should be charged or discharged in each hour. In this paper, we study the real-time V2G control problem under price uncertainty where the electricity price is determined dynamically every hour. Our model is inspired by the Deep Q-learning (DQN) algorithm which combines popular Q-learning with a deep neural network. The proposed Double-DQN model is an update of the DQN which maintains two distinct networks to select or evaluate an action. The Double-DQN algorithm is used to control charge/discharge operation in the hourly available electricity price in order to maximize the profit for the EV owner during the whole parking time. Experiment results show that our proposed method can work effectively in the real electricity market and it is able to increase the profit significantly compared with the other state-of-the-art EV charging schemes.","PeriodicalId":382725,"journal":{"name":"2020 11th International Conference on Information and Knowledge Technology (IKT)","volume":"117 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127716686","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 : 2020-12-22DOI: 10.1109/IKT51791.2020.9345610
Romina Etezadi, M. Shamsfard
Question answering systems may find the answers to users' questions from either unstructured texts or structured data such as knowledge graphs. Answering questions using supervised learning approaches including deep learning models need large training datasets. In recent years, some datasets have been presented for the task of Question answering over knowledge graphs, which is the focus of this paper. Although many datasets in English were proposed, there have been a few question answering datasets in Persian. This paper introduces PeCoQ, a dataset for Persian question answering. This dataset contains 10,000 complex questions and answers extracted from the Persian knowledge graph, FarsBase. For each question, the SPARQL query and two paraphrases that were written by linguists are provided as well. There are different types of complexities in the dataset, such as multi-relation, multi-entity, ordinal, and temporal constraints. In this paper, we discuss the dataset's characteristics and describe our methodolozv for buildinz it.
{"title":"PeCoQ: A Dataset for Persian Complex Question Answering over Knowledge Graph","authors":"Romina Etezadi, M. Shamsfard","doi":"10.1109/IKT51791.2020.9345610","DOIUrl":"https://doi.org/10.1109/IKT51791.2020.9345610","url":null,"abstract":"Question answering systems may find the answers to users' questions from either unstructured texts or structured data such as knowledge graphs. Answering questions using supervised learning approaches including deep learning models need large training datasets. In recent years, some datasets have been presented for the task of Question answering over knowledge graphs, which is the focus of this paper. Although many datasets in English were proposed, there have been a few question answering datasets in Persian. This paper introduces PeCoQ, a dataset for Persian question answering. This dataset contains 10,000 complex questions and answers extracted from the Persian knowledge graph, FarsBase. For each question, the SPARQL query and two paraphrases that were written by linguists are provided as well. There are different types of complexities in the dataset, such as multi-relation, multi-entity, ordinal, and temporal constraints. In this paper, we discuss the dataset's characteristics and describe our methodolozv for buildinz it.","PeriodicalId":382725,"journal":{"name":"2020 11th International Conference on Information and Knowledge Technology (IKT)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121258772","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 : 2020-12-22DOI: 10.1109/IKT51791.2020.9345612
Arash Ebrahimnezad, H. Jazayeriy, Faria Nassiri-Mofakham
In any negotiation, one of the most important parts of the negotiator's task is deciding whether or not to accept the opponent's offer. Actually, the most challenging thing is answering this question: which offer and when must be accepted? A wide range of simple to sophisticated acceptance strategies have been proposed: simple acceptance strategies which have the constant threshold value and sophisticated strategies that notice both utility and time in order to determine acceptance thresholds. This study introduces a novel statistical acceptance strategy with considering the similarity between the opponent's offer and our previous offers, which is combined with existing usual acceptance strategies. Experiments show our strategy has advantages against the state-of-the-art acceptance strategies.
{"title":"Statistical Distance-Based Acceptance Strategy for Desirable Offers in Bilateral Automated Negotiation","authors":"Arash Ebrahimnezad, H. Jazayeriy, Faria Nassiri-Mofakham","doi":"10.1109/IKT51791.2020.9345612","DOIUrl":"https://doi.org/10.1109/IKT51791.2020.9345612","url":null,"abstract":"In any negotiation, one of the most important parts of the negotiator's task is deciding whether or not to accept the opponent's offer. Actually, the most challenging thing is answering this question: which offer and when must be accepted? A wide range of simple to sophisticated acceptance strategies have been proposed: simple acceptance strategies which have the constant threshold value and sophisticated strategies that notice both utility and time in order to determine acceptance thresholds. This study introduces a novel statistical acceptance strategy with considering the similarity between the opponent's offer and our previous offers, which is combined with existing usual acceptance strategies. Experiments show our strategy has advantages against the state-of-the-art acceptance strategies.","PeriodicalId":382725,"journal":{"name":"2020 11th International Conference on Information and Knowledge Technology (IKT)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125033790","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 : 2020-09-01DOI: 10.1109/IKT51791.2020.9345620
Negar Nasiri, M. Keyvanpour
In the recently age, the volume of information is growing exponentially. This data can be used in several fields such as business, healthcare, cyber security, etc. Extracting useful knowledge from raw information is an important process. But the challenge in this process is the sensitivity of this information, which has made owners unwilling to share sensitive information. This has led the study of the privacy of data in data mining to be a hot topic today. In our paper, an aim is made to prepare a framework for qualitative analysis of methods. This qualitative framework consists of three main sections: a comprehensive classification of proposed methods, proposed evaluation criteria and their qualitative evaluation. Our main purpose of presenting this framework is 1) systematic introduction of the most important methods of privacy preserving in data mining 2) creating a suitable platform for qualitative comparison of these methods 3) providing the possibility of selecting methods appropriate to the needs of application areas 4) systematic introduction of points Weakness of existing methods as a prerequisite for improving methods of PPDM.
{"title":"Classification and Evaluation of Privacy Preserving Data Mining Methods","authors":"Negar Nasiri, M. Keyvanpour","doi":"10.1109/IKT51791.2020.9345620","DOIUrl":"https://doi.org/10.1109/IKT51791.2020.9345620","url":null,"abstract":"In the recently age, the volume of information is growing exponentially. This data can be used in several fields such as business, healthcare, cyber security, etc. Extracting useful knowledge from raw information is an important process. But the challenge in this process is the sensitivity of this information, which has made owners unwilling to share sensitive information. This has led the study of the privacy of data in data mining to be a hot topic today. In our paper, an aim is made to prepare a framework for qualitative analysis of methods. This qualitative framework consists of three main sections: a comprehensive classification of proposed methods, proposed evaluation criteria and their qualitative evaluation. Our main purpose of presenting this framework is 1) systematic introduction of the most important methods of privacy preserving in data mining 2) creating a suitable platform for qualitative comparison of these methods 3) providing the possibility of selecting methods appropriate to the needs of application areas 4) systematic introduction of points Weakness of existing methods as a prerequisite for improving methods of PPDM.","PeriodicalId":382725,"journal":{"name":"2020 11th International Conference on Information and Knowledge Technology (IKT)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129316013","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 : 2020-04-22DOI: 10.1109/ikt51791.2020.9345631
Majid Asgari-Bidhendi, Farzane Fakhrian, B. Minaei-Bidgoli
Social media users have exponentially increased in recent years, and social media data has become one of the most populated repositories of data in the world. Natural language text is one of the main portions of this data. However, this textual data contains many entities, which increases the ambiguity of the data. Entity linking targets finding entity mentions and linking them to their corresponding entities in an external dataset. Recently, FarsBase has been introduced as the first Persian knowledge graph, containing almost 750,000 entities. In this study, we propose ParsEL, the first unsupervised end-to-end entity linking system specially designed for the Persian language, and utilizes contextual and graph-based features to rank the candidate entities. To evaluate the proposed approach, we publish the first entity linking dataset for the Persian language, created by crawling social media text from some popular Telegram channels and contains 67,595 tokens. The results show ParsEL records 86.94% f-score for the introduced dataset, and it is comparable with one other entity linking system which supports the Persian language.
{"title":"ParsEL 1.0: Unsupervised Entity Linking in Persian Social Media Texts","authors":"Majid Asgari-Bidhendi, Farzane Fakhrian, B. Minaei-Bidgoli","doi":"10.1109/ikt51791.2020.9345631","DOIUrl":"https://doi.org/10.1109/ikt51791.2020.9345631","url":null,"abstract":"Social media users have exponentially increased in recent years, and social media data has become one of the most populated repositories of data in the world. Natural language text is one of the main portions of this data. However, this textual data contains many entities, which increases the ambiguity of the data. Entity linking targets finding entity mentions and linking them to their corresponding entities in an external dataset. Recently, FarsBase has been introduced as the first Persian knowledge graph, containing almost 750,000 entities. In this study, we propose ParsEL, the first unsupervised end-to-end entity linking system specially designed for the Persian language, and utilizes contextual and graph-based features to rank the candidate entities. To evaluate the proposed approach, we publish the first entity linking dataset for the Persian language, created by crawling social media text from some popular Telegram channels and contains 67,595 tokens. The results show ParsEL records 86.94% f-score for the introduced dataset, and it is comparable with one other entity linking system which supports the Persian language.","PeriodicalId":382725,"journal":{"name":"2020 11th International Conference on Information and Knowledge Technology (IKT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131021260","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：