Pub Date : 2022-11-01DOI: 10.1109/ICA55837.2022.00012
Muhammad Shahroze Ali, F. Azam, Aon Safdar, Muhammad Waseem Anwar
Artificial intelligence (AI)-based chatbot systems have seen increased adaption in the educational domain in recent years owing to increased sophistication in the AI domain. However, most of the communication between students and educational institutions is still performed physically and causes major administrative overhead, especially during the time of admission. Contemporary pattern-matching-based and generative-based chatbots underperform to queries outside a limited scope, grammatically and structurally ambiguous inputs, outliers to pre-defined rule-set, and longer response times for a huge knowledge base. We proposed a NEdBOT-An NLP-based Educational Bot, developed by Natural Language Processing models integrated within the DialogFlow platform utilizing a Retrieval-based approach. We evaluate the developed chatbot on a custom dataset generated for the admissions use case of a prominent university. We used an objective evaluation criterion with real-world users to achieve an intent classification accuracy of 76.8% at an average mean response time of 216.43ms per query and a user-friendliness score of 72% on the System Usability Scale (SUS). The results demonstrate the proposed approach's ability to create robust, reliable, responsive, and user-friendly web-based smart chatbots that are highly scalable with the capability to handle wider scopes and vague inputs with ease.
{"title":"Intelligent Agents in Educational Institutions: NEdBOT - NLP-based Chatbot for Administrative Support Using DialogFlow","authors":"Muhammad Shahroze Ali, F. Azam, Aon Safdar, Muhammad Waseem Anwar","doi":"10.1109/ICA55837.2022.00012","DOIUrl":"https://doi.org/10.1109/ICA55837.2022.00012","url":null,"abstract":"Artificial intelligence (AI)-based chatbot systems have seen increased adaption in the educational domain in recent years owing to increased sophistication in the AI domain. However, most of the communication between students and educational institutions is still performed physically and causes major administrative overhead, especially during the time of admission. Contemporary pattern-matching-based and generative-based chatbots underperform to queries outside a limited scope, grammatically and structurally ambiguous inputs, outliers to pre-defined rule-set, and longer response times for a huge knowledge base. We proposed a NEdBOT-An NLP-based Educational Bot, developed by Natural Language Processing models integrated within the DialogFlow platform utilizing a Retrieval-based approach. We evaluate the developed chatbot on a custom dataset generated for the admissions use case of a prominent university. We used an objective evaluation criterion with real-world users to achieve an intent classification accuracy of 76.8% at an average mean response time of 216.43ms per query and a user-friendliness score of 72% on the System Usability Scale (SUS). The results demonstrate the proposed approach's ability to create robust, reliable, responsive, and user-friendly web-based smart chatbots that are highly scalable with the capability to handle wider scopes and vague inputs with ease.","PeriodicalId":150818,"journal":{"name":"2022 IEEE International Conference on Agents (ICA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122155948","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 : 2022-11-01DOI: 10.1109/ICA55837.2022.00016
Zijia Niu, Yuxin Jin, Wang Yao, Xiao Zhang, Lu Ren
Trajectory planning of massive unmanned aerial vehicles (UAVs) is very difficult in an environment with static and dynamic obstacles. This is mainly due to the huge number of UAVs, which pose challenges to their interaction and collision avoidance with companions and obstacles. In this paper, we propose a trajectory planning algorithm for a massive number of UAVs based on the mean field game (MFG). First, a differential game of N UAVs in a 3D environment is constructed, and the collision avoidance with static and dynamic obstacles is considered in the cost functional of each UAV. Then, when the number of UAVs is very large, the above differential game is transformed into a MFG using the mean field approximation. The existence and uniqueness of the equilibrium solution are proved. Finally, we derive the variational primal-dual formulation of the proposed MFG model and solve it with APAC-Net. The performance of the proposed algorithm is validated in an environment with multiple static obstacles and two different types of dynamic obstacles.
{"title":"Trajectory Planning for A Massive Number of UAVs in the Environment with Static and Dynamic Obstacles: A Mean Field Game Approach","authors":"Zijia Niu, Yuxin Jin, Wang Yao, Xiao Zhang, Lu Ren","doi":"10.1109/ICA55837.2022.00016","DOIUrl":"https://doi.org/10.1109/ICA55837.2022.00016","url":null,"abstract":"Trajectory planning of massive unmanned aerial vehicles (UAVs) is very difficult in an environment with static and dynamic obstacles. This is mainly due to the huge number of UAVs, which pose challenges to their interaction and collision avoidance with companions and obstacles. In this paper, we propose a trajectory planning algorithm for a massive number of UAVs based on the mean field game (MFG). First, a differential game of N UAVs in a 3D environment is constructed, and the collision avoidance with static and dynamic obstacles is considered in the cost functional of each UAV. Then, when the number of UAVs is very large, the above differential game is transformed into a MFG using the mean field approximation. The existence and uniqueness of the equilibrium solution are proved. Finally, we derive the variational primal-dual formulation of the proposed MFG model and solve it with APAC-Net. The performance of the proposed algorithm is validated in an environment with multiple static obstacles and two different types of dynamic obstacles.","PeriodicalId":150818,"journal":{"name":"2022 IEEE International Conference on Agents (ICA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116088887","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 : 2022-11-01DOI: 10.1109/ICA55837.2022.00009
Masanori Hirano, K. Izumi
This study focuses on parameter tuning of artificial market simulations and aims to replace the traditional qualitative evaluation metrics based on stylized facts with the proposed quantitative metrics. Traditionally, for the evaluation of artificial market simulations, the replication of stylized facts, a common phenomenon among financial markets and observed in empirical studies, is verified by humans. However, this prevents large-scale parameter tuning owing to the complexity of automation. Hence, this study utilizes a generative adversarial network (GAN) for this replacement because we assume that the GAN's learning architecture has a good fit for evaluating the distributional features of actual markets and can learn stylized facts implicitly. In the proposed parameter-tuning method, the simulated data are input into the critic of the GAN, and the outputs are employed as the objective value of the tuning. The parameter tuning results show that we successfully tuned the high-dimensional parameters of artificial market simulations and confirmed that the optimized parameter could replicate the stylized facts employed in traditional qualitative evaluation metrics.
{"title":"Quantitative Tuning of Artificial Market Simulation using Generative Adversarial Network","authors":"Masanori Hirano, K. Izumi","doi":"10.1109/ICA55837.2022.00009","DOIUrl":"https://doi.org/10.1109/ICA55837.2022.00009","url":null,"abstract":"This study focuses on parameter tuning of artificial market simulations and aims to replace the traditional qualitative evaluation metrics based on stylized facts with the proposed quantitative metrics. Traditionally, for the evaluation of artificial market simulations, the replication of stylized facts, a common phenomenon among financial markets and observed in empirical studies, is verified by humans. However, this prevents large-scale parameter tuning owing to the complexity of automation. Hence, this study utilizes a generative adversarial network (GAN) for this replacement because we assume that the GAN's learning architecture has a good fit for evaluating the distributional features of actual markets and can learn stylized facts implicitly. In the proposed parameter-tuning method, the simulated data are input into the critic of the GAN, and the outputs are employed as the objective value of the tuning. The parameter tuning results show that we successfully tuned the high-dimensional parameters of artificial market simulations and confirmed that the optimized parameter could replicate the stylized facts employed in traditional qualitative evaluation metrics.","PeriodicalId":150818,"journal":{"name":"2022 IEEE International Conference on Agents (ICA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132782479","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 : 2022-11-01DOI: 10.1109/ICA55837.2022.00014
Hae-Na Kim, Sangho Lee, Jiyoung Woo, H. Kim
When rank and skill do not coincide in competitive games, this might be a sign of issues such as boosting, smurfing, and trolling occur. The fair gaming culture of online gaming is disrupted and offended by cheating like boosting, smurfing, and trolling. The player's play style must be used to determine the rank that appears in account information. In this study, we classified League of Legends' low and high tiers using a sequence-based CNN-LSTM model. Using input perturbation, the model can explain its own importance for certain features. The experimental progress: First, we selected features that show a difference between tiers by an extracted score estimating a cumulative sum graph. Second, we construct the dataset format variously with variable or fixed sequence length, compare performance, and analyze the pros and cons. Finally, we consider the possibility of early detection by measuring performance over game elapsed time. Along with the experiment, a rank classification performance of the model achieved AUC 0.9036 and found that we can distinguish from the 24 minutes after the start of the game. In addition, We derived that ccReduction and MinionsKilied were the information that had the most influence on skills among various features.
{"title":"Justice League: Time-series Game Player Pattern Detection to Discover Rank-Skill Mismatch","authors":"Hae-Na Kim, Sangho Lee, Jiyoung Woo, H. Kim","doi":"10.1109/ICA55837.2022.00014","DOIUrl":"https://doi.org/10.1109/ICA55837.2022.00014","url":null,"abstract":"When rank and skill do not coincide in competitive games, this might be a sign of issues such as boosting, smurfing, and trolling occur. The fair gaming culture of online gaming is disrupted and offended by cheating like boosting, smurfing, and trolling. The player's play style must be used to determine the rank that appears in account information. In this study, we classified League of Legends' low and high tiers using a sequence-based CNN-LSTM model. Using input perturbation, the model can explain its own importance for certain features. The experimental progress: First, we selected features that show a difference between tiers by an extracted score estimating a cumulative sum graph. Second, we construct the dataset format variously with variable or fixed sequence length, compare performance, and analyze the pros and cons. Finally, we consider the possibility of early detection by measuring performance over game elapsed time. Along with the experiment, a rank classification performance of the model achieved AUC 0.9036 and found that we can distinguish from the 24 minutes after the start of the game. In addition, We derived that ccReduction and MinionsKilied were the information that had the most influence on skills among various features.","PeriodicalId":150818,"journal":{"name":"2022 IEEE International Conference on Agents (ICA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133156982","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 : 2022-11-01DOI: 10.1109/ICA55837.2022.00008
Lui Cirocco, D. Jarvis, J. Jarvis, R. Rönnquist
Behavior trees are becoming increasingly used in both the computer games and robotics industries as a mechanism for non-programmers to specify individual entity (non-playing character or robot) behaviors. The tree representation that is employed is amenable to graphical representation and the availability of graphical editors has contributed significantly to the uptake of the approach. Working with graphical representations of behavior makes development by non-programmers easier. However, existing behavior tree representations are limited in their ability to represent team behavior and complex entity reasoning. Agent frameworks exist that address these issues and while they have been successfully employed in military war gaming applications, the underlying behavior representation that they use are fundamentally different to behavior trees. In this paper, we introduce GORITE, a BDI agent framework in which agent behavior is specified using goal-based process models which are representationally similar to behavior trees. Unlike behavior trees, process models can be used to represent both individual and team behaviors. Furthermore, GORITE provides support to enable an agent (or team of agents) to reason about the goals that it intends to pursue as well as the goals that it is currently pursuing. A simple example is used to demonstrate how team behavior can be modelled using GORITE process models.
{"title":"GORITE: A BDI Realisation of Behavior Trees","authors":"Lui Cirocco, D. Jarvis, J. Jarvis, R. Rönnquist","doi":"10.1109/ICA55837.2022.00008","DOIUrl":"https://doi.org/10.1109/ICA55837.2022.00008","url":null,"abstract":"Behavior trees are becoming increasingly used in both the computer games and robotics industries as a mechanism for non-programmers to specify individual entity (non-playing character or robot) behaviors. The tree representation that is employed is amenable to graphical representation and the availability of graphical editors has contributed significantly to the uptake of the approach. Working with graphical representations of behavior makes development by non-programmers easier. However, existing behavior tree representations are limited in their ability to represent team behavior and complex entity reasoning. Agent frameworks exist that address these issues and while they have been successfully employed in military war gaming applications, the underlying behavior representation that they use are fundamentally different to behavior trees. In this paper, we introduce GORITE, a BDI agent framework in which agent behavior is specified using goal-based process models which are representationally similar to behavior trees. Unlike behavior trees, process models can be used to represent both individual and team behaviors. Furthermore, GORITE provides support to enable an agent (or team of agents) to reason about the goals that it intends to pursue as well as the goals that it is currently pursuing. A simple example is used to demonstrate how team behavior can be modelled using GORITE process models.","PeriodicalId":150818,"journal":{"name":"2022 IEEE International Conference on Agents (ICA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"113998869","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 : 2022-11-01DOI: 10.1109/ICA55837.2022.00017
Jie Jiang, P. Long, Lijia Xie, Z. Zheng
Wireless Sensor Networks (WSN) have assisted applications of multi-agent system. Abundant sensor nodes, densely distributed around a base station (BS), collect data and transmit to BS node for data analysis. The concept of cluster has been emerged as the efficient communication structure in resource-constrained environment. However, the security still remains a major concern due to the vulnerability of sensor nodes. In this paper, we propose a percolation-based secure routing protocol. We leverage the trust score composed of three indexes to select cluster heads (CH) for unevenly distributed clusters. By considering the reliability, centrality and stability, legitimate nodes with social trust and adequate energy are chosen to provide relay service. Moreover, we design a multi-path inter-cluster routing protocol to construct CH chains for directed inter-cluster data transmission based on the percolation. And the measurement of transit score for on-path CH nodes contributes to load balancing and security. Our simulation results show that our protocol is able to guarantee the security to improve the delivery ratio and packets delay.
{"title":"A Percolation-Based Secure Routing Protocol for Wireless Sensor Networks","authors":"Jie Jiang, P. Long, Lijia Xie, Z. Zheng","doi":"10.1109/ICA55837.2022.00017","DOIUrl":"https://doi.org/10.1109/ICA55837.2022.00017","url":null,"abstract":"Wireless Sensor Networks (WSN) have assisted applications of multi-agent system. Abundant sensor nodes, densely distributed around a base station (BS), collect data and transmit to BS node for data analysis. The concept of cluster has been emerged as the efficient communication structure in resource-constrained environment. However, the security still remains a major concern due to the vulnerability of sensor nodes. In this paper, we propose a percolation-based secure routing protocol. We leverage the trust score composed of three indexes to select cluster heads (CH) for unevenly distributed clusters. By considering the reliability, centrality and stability, legitimate nodes with social trust and adequate energy are chosen to provide relay service. Moreover, we design a multi-path inter-cluster routing protocol to construct CH chains for directed inter-cluster data transmission based on the percolation. And the measurement of transit score for on-path CH nodes contributes to load balancing and security. Our simulation results show that our protocol is able to guarantee the security to improve the delivery ratio and packets delay.","PeriodicalId":150818,"journal":{"name":"2022 IEEE International Conference on Agents (ICA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125557206","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 : 2022-11-01DOI: 10.1109/ICA55837.2022.00013
Ruhollah Jamali, S. Lazarova-Molnar
Pharmaceutical parallel trade emerged due to the European Union's single market for medicines. While many players, such as manufacturers, wholesalers, parallel traders, pharmacies, regulatory authorities, and hospitals, are involved in this market, having a model that accurately reflects the parallel trade market could be a considerable advantage for players in this market. One way to model the parallel trade market is by employing game theory, which is frequently used to model and explain business interactions. However, game theory imposes limitations on models. Agent-based modeling is a promising framework for studying the parallel trade market, which allows us to investigate macroscopic outcomes that emerge from microscopic rules, decisions, and interactions. Moreover, agent-based modeling allows for high expressiveness and complexity in agents, improving agents' efficiency in autonomy and reactivity compared to current game theoretic models. In this paper, we develop the concept of an agent-based model for the pharmaceutical parallel trading market based on an available game-theoretic model of the market.
{"title":"The Relationship Between Agent-based Simulation and Game Theory in the Case of Parallel Trade","authors":"Ruhollah Jamali, S. Lazarova-Molnar","doi":"10.1109/ICA55837.2022.00013","DOIUrl":"https://doi.org/10.1109/ICA55837.2022.00013","url":null,"abstract":"Pharmaceutical parallel trade emerged due to the European Union's single market for medicines. While many players, such as manufacturers, wholesalers, parallel traders, pharmacies, regulatory authorities, and hospitals, are involved in this market, having a model that accurately reflects the parallel trade market could be a considerable advantage for players in this market. One way to model the parallel trade market is by employing game theory, which is frequently used to model and explain business interactions. However, game theory imposes limitations on models. Agent-based modeling is a promising framework for studying the parallel trade market, which allows us to investigate macroscopic outcomes that emerge from microscopic rules, decisions, and interactions. Moreover, agent-based modeling allows for high expressiveness and complexity in agents, improving agents' efficiency in autonomy and reactivity compared to current game theoretic models. In this paper, we develop the concept of an agent-based model for the pharmaceutical parallel trading market based on an available game-theoretic model of the market.","PeriodicalId":150818,"journal":{"name":"2022 IEEE International Conference on Agents (ICA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128441485","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 : 2022-11-01DOI: 10.1109/ICA55837.2022.00018
Warisara Asawaponwiput, Panyawut Sriiesaranusorn, Thawat Mohchit, N. Thatphithakkul, D. Surangsrirat
Nowadays, people are increasingly concerned for their health as being healthy is regarded as a profitable investment. Obesity is one of the most common health problems that leads to multiple diseases. We work with the team that developed a mobile application to encourage users to change their eating and activity behaviors to improve their health based on a virtual competition platform. Participants are required to upload a weight-in photo to verify their weight before the challenge. Manually verifying these images can be time-consuming and error-prone due to the large number of images in each competition. In this study, we proposed an image classification approach to help screen incorrect images of the weight-in photo for the virtual competition. The image augmentation techniques were applied to the training images before being input into the classification model. Since the goal is to deploy the model in a mobile application, the suitable model must be small and efficient enough for use in a limited resources environment. Therefore, VGGNet-16 and MobileNet-V2 were selected as the classification models. The experimental results show that the model could learn from the preprocessed images and obtain satisfactory classification results from pre-trained VGGNet-16 with the highest accuracy and F1-score of 95.00% and 95.23%, respectively. MobileNet-V2 inference time was approximately 10 times faster but the performance was lower with the highest accuracy and F1-score of 93.00% and 93.32%, respectively.
{"title":"Application of Machine Learning in Lifestyle: Weight-In Image Classification using Convolutional Neural Networks","authors":"Warisara Asawaponwiput, Panyawut Sriiesaranusorn, Thawat Mohchit, N. Thatphithakkul, D. Surangsrirat","doi":"10.1109/ICA55837.2022.00018","DOIUrl":"https://doi.org/10.1109/ICA55837.2022.00018","url":null,"abstract":"Nowadays, people are increasingly concerned for their health as being healthy is regarded as a profitable investment. Obesity is one of the most common health problems that leads to multiple diseases. We work with the team that developed a mobile application to encourage users to change their eating and activity behaviors to improve their health based on a virtual competition platform. Participants are required to upload a weight-in photo to verify their weight before the challenge. Manually verifying these images can be time-consuming and error-prone due to the large number of images in each competition. In this study, we proposed an image classification approach to help screen incorrect images of the weight-in photo for the virtual competition. The image augmentation techniques were applied to the training images before being input into the classification model. Since the goal is to deploy the model in a mobile application, the suitable model must be small and efficient enough for use in a limited resources environment. Therefore, VGGNet-16 and MobileNet-V2 were selected as the classification models. The experimental results show that the model could learn from the preprocessed images and obtain satisfactory classification results from pre-trained VGGNet-16 with the highest accuracy and F1-score of 95.00% and 95.23%, respectively. MobileNet-V2 inference time was approximately 10 times faster but the performance was lower with the highest accuracy and F1-score of 93.00% and 93.32%, respectively.","PeriodicalId":150818,"journal":{"name":"2022 IEEE International Conference on Agents (ICA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129848335","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 : 2022-11-01DOI: 10.1109/ICA55837.2022.00010
Hikaru Ishizuka, Shun Shiramatsu, Keiko Ono
The term “sublation” (or “aufheben”) refers to the process of arriving at an agreed upon answer to two opposing arguments without denying either of them. In this study, we conducted a discussion experiment in which we quantified the degree of sublation and analyzed the results to determine the factors that contribute to the cessation of conflicting opinions in discussions. Our findings revealed a weak positive correlation between the number of URLs posted as evidence for one's opinion and the degree of sublation of the consensus proposal. In actual discussions and debates, however, there are times when everyone makes the same argument, with little or no opposing views, resulting in biased opinions. To address this problem, we developed a method to eliminate bias in opinions, in which an agent posts information that reinforces the opinion of a minority in a discussion. The experimental results demonstrate that GPT-3, a natural language processing model, can be applied to summarization of relevant information for information provision and the resolution of opinion bias.
{"title":"Prototyping Agents for Resolving Opinion Biases Toward Facilitating Sublation of Conflict in Web-based Discussions","authors":"Hikaru Ishizuka, Shun Shiramatsu, Keiko Ono","doi":"10.1109/ICA55837.2022.00010","DOIUrl":"https://doi.org/10.1109/ICA55837.2022.00010","url":null,"abstract":"The term “sublation” (or “aufheben”) refers to the process of arriving at an agreed upon answer to two opposing arguments without denying either of them. In this study, we conducted a discussion experiment in which we quantified the degree of sublation and analyzed the results to determine the factors that contribute to the cessation of conflicting opinions in discussions. Our findings revealed a weak positive correlation between the number of URLs posted as evidence for one's opinion and the degree of sublation of the consensus proposal. In actual discussions and debates, however, there are times when everyone makes the same argument, with little or no opposing views, resulting in biased opinions. To address this problem, we developed a method to eliminate bias in opinions, in which an agent posts information that reinforces the opinion of a minority in a discussion. The experimental results demonstrate that GPT-3, a natural language processing model, can be applied to summarization of relevant information for information provision and the resolution of opinion bias.","PeriodicalId":150818,"journal":{"name":"2022 IEEE International Conference on Agents (ICA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131111187","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 : 2022-11-01DOI: 10.1109/ICA55837.2022.00011
Ryosuke Kinoshita, Shun Shiramatsu
In Web discussions, which have become mainstream with COVID-19, the amount of information possessed and the level of understanding of the discussion differ among participants. As a result, some participants may not be able to speak up satisfactorily, and this can hinder consensus building in the discussion as a whole. Therefore, we develop an agent that automatically recommends information related to the discussion as information that facilitates participants to speak up. The agent first obtains necessary discussion data from on-going Web discussions. The information to be recommended is determined by real-time search. Query words for the search are generated using a pre-trained query-term-generation model. When selecting information to recommend from the information obtained in the search, a model that classifies the acquired information according to the discussion phase is used. The results of a discussion experiment in which an agent intervened in a Web-based discussion showed many results indicating the effectiveness of the agent, although there are some points that need to be improved. However, since the scale of the discussion experiment was small, it will be necessary to validate the agent in large-scale discussions in the future.
{"title":"Agent for Recommending Information Relevant to Web-based Discussion by Generating Query Terms using GPT-3","authors":"Ryosuke Kinoshita, Shun Shiramatsu","doi":"10.1109/ICA55837.2022.00011","DOIUrl":"https://doi.org/10.1109/ICA55837.2022.00011","url":null,"abstract":"In Web discussions, which have become mainstream with COVID-19, the amount of information possessed and the level of understanding of the discussion differ among participants. As a result, some participants may not be able to speak up satisfactorily, and this can hinder consensus building in the discussion as a whole. Therefore, we develop an agent that automatically recommends information related to the discussion as information that facilitates participants to speak up. The agent first obtains necessary discussion data from on-going Web discussions. The information to be recommended is determined by real-time search. Query words for the search are generated using a pre-trained query-term-generation model. When selecting information to recommend from the information obtained in the search, a model that classifies the acquired information according to the discussion phase is used. The results of a discussion experiment in which an agent intervened in a Web-based discussion showed many results indicating the effectiveness of the agent, although there are some points that need to be improved. However, since the scale of the discussion experiment was small, it will be necessary to validate the agent in large-scale discussions in the future.","PeriodicalId":150818,"journal":{"name":"2022 IEEE International Conference on Agents (ICA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121557158","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：