Pub Date : 2021-11-01DOI: 10.1109/taai54685.2021.00012
Yen-Hao Huang, Tzu-Yun Lee, Fernando H. Calderon, Yi-Shin Chen
Span selection is an important prerequisite for many natural language processing tasks. Existing methods usually generate phrase-like spans from entire articles without leveraging the topics or the key points within each paragraph that usually lie behind sentence generation during the writing processes. This study looks at multi-sentence span selection for generating multiple, independent, key-point spans with complete endings for news articles. The proposed span selection model consists of a context relation model and an end span model that merge context-related sentences within a span. The context relation model captures the topics shared between sentences, and the end span model utilizes the embeddings of Zhuyin, the orthography of Mandarin, and the cross attention between words and Zhuyin to effectively capture the end positions of the spans. To evaluate the proposed framework, we construct a news report dataset in Mandarin. Experimental results show that the proposed model not only improves performance, but is also better than previous approaches and close to human span production. The proposed Zhuyin embeddings and cross-attention also improve on BERT’s end sentence detection performance in Mandarin.
{"title":"Dynamic Span Selection for Mandarin Articles Using Contextual Relations and Orthography","authors":"Yen-Hao Huang, Tzu-Yun Lee, Fernando H. Calderon, Yi-Shin Chen","doi":"10.1109/taai54685.2021.00012","DOIUrl":"https://doi.org/10.1109/taai54685.2021.00012","url":null,"abstract":"Span selection is an important prerequisite for many natural language processing tasks. Existing methods usually generate phrase-like spans from entire articles without leveraging the topics or the key points within each paragraph that usually lie behind sentence generation during the writing processes. This study looks at multi-sentence span selection for generating multiple, independent, key-point spans with complete endings for news articles. The proposed span selection model consists of a context relation model and an end span model that merge context-related sentences within a span. The context relation model captures the topics shared between sentences, and the end span model utilizes the embeddings of Zhuyin, the orthography of Mandarin, and the cross attention between words and Zhuyin to effectively capture the end positions of the spans. To evaluate the proposed framework, we construct a news report dataset in Mandarin. Experimental results show that the proposed model not only improves performance, but is also better than previous approaches and close to human span production. The proposed Zhuyin embeddings and cross-attention also improve on BERT’s end sentence detection performance in Mandarin.","PeriodicalId":343821,"journal":{"name":"2021 International Conference on Technologies and Applications of Artificial Intelligence (TAAI)","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116480171","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-01DOI: 10.1109/taai54685.2021.00034
J. Sum, Janet C.C. Chang
Recently, it has been shown that gradient descent learning (GDL) might have problem in training a neural network (NN) with persistent weight noise. In the presence of multiplicative weight noise (resp. node noise), the model generated by GDL is not the desired model which minimizes the expected mean-squared-error (MSE) subjected to multiplicative weight noise (resp. node noise). In this paper, the analysis is formalized under a conceptual framework called suitability and extended to the learning gradient descent with momentum (GDM). A learning algorithm is suitable to be implemented on-chip to train a NN with weight noise if its learning objective is identical to the expected MSE of the NN with the same noise. In this regard, it is shown that GDL and GDM are not suitable to be implemented on-chip. Theoretical analysis in support with experimental evidences are presented for the claims.
{"title":"On Gradient Descent for On-Chip Learning","authors":"J. Sum, Janet C.C. Chang","doi":"10.1109/taai54685.2021.00034","DOIUrl":"https://doi.org/10.1109/taai54685.2021.00034","url":null,"abstract":"Recently, it has been shown that gradient descent learning (GDL) might have problem in training a neural network (NN) with persistent weight noise. In the presence of multiplicative weight noise (resp. node noise), the model generated by GDL is not the desired model which minimizes the expected mean-squared-error (MSE) subjected to multiplicative weight noise (resp. node noise). In this paper, the analysis is formalized under a conceptual framework called suitability and extended to the learning gradient descent with momentum (GDM). A learning algorithm is suitable to be implemented on-chip to train a NN with weight noise if its learning objective is identical to the expected MSE of the NN with the same noise. In this regard, it is shown that GDL and GDM are not suitable to be implemented on-chip. Theoretical analysis in support with experimental evidences are presented for the claims.","PeriodicalId":343821,"journal":{"name":"2021 International Conference on Technologies and Applications of Artificial Intelligence (TAAI)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132430918","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-01DOI: 10.1109/taai54685.2021.00001
{"title":"[Title page i]","authors":"","doi":"10.1109/taai54685.2021.00001","DOIUrl":"https://doi.org/10.1109/taai54685.2021.00001","url":null,"abstract":"","PeriodicalId":343821,"journal":{"name":"2021 International Conference on Technologies and Applications of Artificial Intelligence (TAAI)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129037682","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-01DOI: 10.1109/taai54685.2021.00059
Thao-Trang Huynh-Cam, Zi-Jie Luo, Long-Sheng Chen
In recent years, live streaming has developed rapidly in the world and become one of the most popular entertainment activities of most people since 2011, especially the youth due to the rich and various content. Previous literatures mainly focused on finding popular streamer and behaviors of live streaming viewers like gift giving behaviors. However, the studies on the effect of reviewers’ comments on the number of viewing and on text comments via live chat rooms of social media users, especially of games live streaming users are very limited, although these issues are considered to significantly and positively affect others’ behaviors. Therefore, this work aims to use the text comments in the live chat room as input variables to predict the number of viewing games live streaming. Random forests (RF) and decision trees (DT) algorithms were employed to build prediction models. Game live streaming was our research target. The prediction accuracy rate of the established model is nearly 90%. The analysis results is expected to be a roadmap for the live streaming platforms to carefully respond viewers’ comments.
{"title":"Using Random Forests and Decision Trees to Predict Viewing Game Live Streaming via Viewers’ Comments","authors":"Thao-Trang Huynh-Cam, Zi-Jie Luo, Long-Sheng Chen","doi":"10.1109/taai54685.2021.00059","DOIUrl":"https://doi.org/10.1109/taai54685.2021.00059","url":null,"abstract":"In recent years, live streaming has developed rapidly in the world and become one of the most popular entertainment activities of most people since 2011, especially the youth due to the rich and various content. Previous literatures mainly focused on finding popular streamer and behaviors of live streaming viewers like gift giving behaviors. However, the studies on the effect of reviewers’ comments on the number of viewing and on text comments via live chat rooms of social media users, especially of games live streaming users are very limited, although these issues are considered to significantly and positively affect others’ behaviors. Therefore, this work aims to use the text comments in the live chat room as input variables to predict the number of viewing games live streaming. Random forests (RF) and decision trees (DT) algorithms were employed to build prediction models. Game live streaming was our research target. The prediction accuracy rate of the established model is nearly 90%. The analysis results is expected to be a roadmap for the live streaming platforms to carefully respond viewers’ comments.","PeriodicalId":343821,"journal":{"name":"2021 International Conference on Technologies and Applications of Artificial Intelligence (TAAI)","volume":"395 3","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"113997255","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-01DOI: 10.1109/taai54685.2021.00022
Min-Te Sun, Kotcharat Kitchat, Li-Chung Hsieh
Traditional research on the stock and/or futures price prediction mostly uses the past stock/future prices and technique indicators, such as KD, RSI, and MACD, as features. Very few studies consider the forum messages or bargaining chips as stock and/or futures price prediction features. In this research, discussion messages from both PTT and CMoney forums are converted into daily sentimental vectors using the retrained BERT. The daily sentimental vector as well as three bargaining chips are then used as features to train the GRU and TCN models. The experiment results show that the TCN performs better than the GRU-based RNN model in terms of MAE, MAPE, RMSE, and accuracy. In addition, both of the bargaining chips and forum messages are verified to be useful in the futures price prediction. The market simulations based on the historical futures price show that a simple investment strategy using the TCN models with techniques, bargaining chips, and forum messages can earn more than 7 times of the investment in the period of one year.
{"title":"TCN-based Futures Prediction Using Financial Indices, Bargain Chips, and Forum Messages","authors":"Min-Te Sun, Kotcharat Kitchat, Li-Chung Hsieh","doi":"10.1109/taai54685.2021.00022","DOIUrl":"https://doi.org/10.1109/taai54685.2021.00022","url":null,"abstract":"Traditional research on the stock and/or futures price prediction mostly uses the past stock/future prices and technique indicators, such as KD, RSI, and MACD, as features. Very few studies consider the forum messages or bargaining chips as stock and/or futures price prediction features. In this research, discussion messages from both PTT and CMoney forums are converted into daily sentimental vectors using the retrained BERT. The daily sentimental vector as well as three bargaining chips are then used as features to train the GRU and TCN models. The experiment results show that the TCN performs better than the GRU-based RNN model in terms of MAE, MAPE, RMSE, and accuracy. In addition, both of the bargaining chips and forum messages are verified to be useful in the futures price prediction. The market simulations based on the historical futures price show that a simple investment strategy using the TCN models with techniques, bargaining chips, and forum messages can earn more than 7 times of the investment in the period of one year.","PeriodicalId":343821,"journal":{"name":"2021 International Conference on Technologies and Applications of Artificial Intelligence (TAAI)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129183198","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}
Grasping objects is one of the basic functions of a robot arm. This study completed the implementation of the process in which a collaborative robot (cobot) arm grasps an object. Hardware components included a depth camera, cobot arm, and artificial intelligence equipment for edge computing. Software components included computer visualization techniques, deep learning, and robot operating system. To complete the preliminary implementation of the system, the grasping operation of the robot arm was set to target a ball. This system implementation sheds light on how robot arms and deep learning techniques are applied to real-life problems. Experiments verified that the preliminary system implemented was able to correctly complete the ball-grasping operation and achieve the pragmatic goal.
{"title":"Preliminary Implementation of Grasping Operation by a Collaborative Robot Arm: Using a Ball as Example","authors":"Wen-Chang Cheng, Chien-Hung Lin, Cheng-Yi Shi, Hung-Chou Hsiao, Chun-Lung Chang","doi":"10.1109/taai54685.2021.00062","DOIUrl":"https://doi.org/10.1109/taai54685.2021.00062","url":null,"abstract":"Grasping objects is one of the basic functions of a robot arm. This study completed the implementation of the process in which a collaborative robot (cobot) arm grasps an object. Hardware components included a depth camera, cobot arm, and artificial intelligence equipment for edge computing. Software components included computer visualization techniques, deep learning, and robot operating system. To complete the preliminary implementation of the system, the grasping operation of the robot arm was set to target a ball. This system implementation sheds light on how robot arms and deep learning techniques are applied to real-life problems. Experiments verified that the preliminary system implemented was able to correctly complete the ball-grasping operation and achieve the pragmatic goal.","PeriodicalId":343821,"journal":{"name":"2021 International Conference on Technologies and Applications of Artificial Intelligence (TAAI)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127551222","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-01DOI: 10.1109/taai54685.2021.00042
Jeang-Kuo Chen, Wei-Zhe Lee
In the past, Relational Database (RDB) is widely used to access data quickly because of high data accuracy and consistency. However, the Big Data era forces many industries to change RDB in order to process a huge amount of data in time. NoSQL (Not only SQL) database (NoSQL DB) is a tailor-made DB. Various industries can choose a suitable NoSQL DB to enhance their competitiveness according to their operational needs. But the data in RDB still need to be saved for company’s operation. How to convert data from RDB to NoSQL DB is an important issue. This paper proposes an algorithm to transfer an RDB to the Wide Column Store DB (WCSDB), one of the most popular NoSQL databases. The experiment results show that WCSDB performs better than RDB.
在过去，关系型数据库(RDB)由于具有较高的数据准确性和一致性，被广泛用于快速访问数据。然而，大数据时代迫使许多行业改变RDB，以便及时处理大量数据。NoSQL (Not only SQL)数据库(NoSQL DB)是一种量身定制的数据库。各行业可以根据自身的运营需要，选择合适的NoSQL数据库来提升自身的竞争力。但是RDB中的数据仍然需要保存，以供公司运营使用。如何将数据从RDB转换为NoSQL数据库是一个重要的问题。本文提出了一种将RDB传输到最流行的NoSQL数据库之一宽列存储数据库(WCSDB)的算法。实验结果表明，WCSDB的性能优于RDB。
{"title":"The Transformation of RDB to NoSQL DB","authors":"Jeang-Kuo Chen, Wei-Zhe Lee","doi":"10.1109/taai54685.2021.00042","DOIUrl":"https://doi.org/10.1109/taai54685.2021.00042","url":null,"abstract":"In the past, Relational Database (RDB) is widely used to access data quickly because of high data accuracy and consistency. However, the Big Data era forces many industries to change RDB in order to process a huge amount of data in time. NoSQL (Not only SQL) database (NoSQL DB) is a tailor-made DB. Various industries can choose a suitable NoSQL DB to enhance their competitiveness according to their operational needs. But the data in RDB still need to be saved for company’s operation. How to convert data from RDB to NoSQL DB is an important issue. This paper proposes an algorithm to transfer an RDB to the Wide Column Store DB (WCSDB), one of the most popular NoSQL databases. The experiment results show that WCSDB performs better than RDB.","PeriodicalId":343821,"journal":{"name":"2021 International Conference on Technologies and Applications of Artificial Intelligence (TAAI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131061661","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-01DOI: 10.1109/taai54685.2021.00024
Ruo-Wei Hung, Ming-Jung Chiu
Let G be a graph with vertex set V(G) and edge set E(G). A set ⊆ V(G) is a dominating set of G if every vertex not in D is adjacent to one vertex in D. The domination problem on G is to compute a dominating set of G with minimum cardinality. A set S ⊆ V(G) is called a paired restrained-dominating set of G if S is a dominating set of G, the subgraph induced by S contains a perfect matching, and the subgraph induced by V(G)−S contains no isolated vertex. The paired restrained- domination number γp(G) of a graph G is the minimum size of a paired restrained-dominating set in G. The paired restrained- domination problem on a graph G is to find a paired restrained- dominating set of G with cardinality γp (G), and is first introduced here. Extending supergrid graphs form a superclass of grid graphs, diagonal supergrid graphs, and supergrid graphs. The domination problem for grid graphs was known to be NP-complete and hence it is NP-complete on extending supergrid graphs. In the past, we have proved the domination problem on supergrid graphs to be NP-complete. The complexity of the paired restrained-domination problem on grid, diagonal supergrid, and supergrid graphs is still unknown. In this paper, we will prove it to be NP-complete for diagonal supergrid graphs, and hence it is NP-complete for extending supergrid graphs. This result can be extended to supergrid graphs. We then provide an upper bound of γp (Rm×n) for rectangular supergrid graph Rm×n.
{"title":"The Paired Restrained-Domination Problem in Supergrid Graphs","authors":"Ruo-Wei Hung, Ming-Jung Chiu","doi":"10.1109/taai54685.2021.00024","DOIUrl":"https://doi.org/10.1109/taai54685.2021.00024","url":null,"abstract":"Let G be a graph with vertex set V(G) and edge set E(G). A set ⊆ V(G) is a dominating set of G if every vertex not in D is adjacent to one vertex in D. The domination problem on G is to compute a dominating set of G with minimum cardinality. A set S ⊆ V(G) is called a paired restrained-dominating set of G if S is a dominating set of G, the subgraph induced by S contains a perfect matching, and the subgraph induced by V(G)−S contains no isolated vertex. The paired restrained- domination number γp(G) of a graph G is the minimum size of a paired restrained-dominating set in G. The paired restrained- domination problem on a graph G is to find a paired restrained- dominating set of G with cardinality γp (G), and is first introduced here. Extending supergrid graphs form a superclass of grid graphs, diagonal supergrid graphs, and supergrid graphs. The domination problem for grid graphs was known to be NP-complete and hence it is NP-complete on extending supergrid graphs. In the past, we have proved the domination problem on supergrid graphs to be NP-complete. The complexity of the paired restrained-domination problem on grid, diagonal supergrid, and supergrid graphs is still unknown. In this paper, we will prove it to be NP-complete for diagonal supergrid graphs, and hence it is NP-complete for extending supergrid graphs. This result can be extended to supergrid graphs. We then provide an upper bound of γp (Rm×n) for rectangular supergrid graph Rm×n.","PeriodicalId":343821,"journal":{"name":"2021 International Conference on Technologies and Applications of Artificial Intelligence (TAAI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124394643","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-01DOI: 10.1109/taai54685.2021.00067
C. Chang, Hsin-Ming Tseng, H. Chu
Before the advent of deep learning, traditional image recognition used algorithms to find features and then classify images using classical machine learning algorithms. But it is difficult to define features for variable types of images. Instead, Representation learning is a way to allow a system to discover the representations of feature for image processing. Deep-learning algorithms attempt to learn multiple levels of representation and play the key role of modern Representation learning. When deep learning neural networks recognize images, the shallow layers usually extract lower-level features, then start with intermediate-level features, and finally the full images. This study uses CNN (Convolutional Neural Network) to classify fireworks images, and the trained modules are used to evaluate the accuracy and suitability.
{"title":"Fireworks Image Classification with Deep Learning","authors":"C. Chang, Hsin-Ming Tseng, H. Chu","doi":"10.1109/taai54685.2021.00067","DOIUrl":"https://doi.org/10.1109/taai54685.2021.00067","url":null,"abstract":"Before the advent of deep learning, traditional image recognition used algorithms to find features and then classify images using classical machine learning algorithms. But it is difficult to define features for variable types of images. Instead, Representation learning is a way to allow a system to discover the representations of feature for image processing. Deep-learning algorithms attempt to learn multiple levels of representation and play the key role of modern Representation learning. When deep learning neural networks recognize images, the shallow layers usually extract lower-level features, then start with intermediate-level features, and finally the full images. This study uses CNN (Convolutional Neural Network) to classify fireworks images, and the trained modules are used to evaluate the accuracy and suitability.","PeriodicalId":343821,"journal":{"name":"2021 International Conference on Technologies and Applications of Artificial Intelligence (TAAI)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126722562","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-01DOI: 10.1109/taai54685.2021.00029
Eric Tswen Gwo Wang, Abbott Po Shun Chen, C. Liu
This paper presents the realization of artificial intelligence (AI) in the Agricultural Industry driven by the chatbot impact on user experience. The study implemented past evaluation methods of Retrieval-based and Generative-based AI chatbots, including efficiency and construction costs. determine the extent of the customer satisfaction factors that successfully influence chatbot acceptance in Taiwan. The results of an empirical study demonstrated that an integrated method of retrieval and natural generation is the most suitable for agricultural services. In addition to the user's perception, the chatbot must consider the overall value evaluation, which Hybrid retrieval-generation can also achieve the most appropriate system efficiency and user satisfaction.
{"title":"A Hybrid Evaluation of AI Chatbots in Taiwan Agriculture Services","authors":"Eric Tswen Gwo Wang, Abbott Po Shun Chen, C. Liu","doi":"10.1109/taai54685.2021.00029","DOIUrl":"https://doi.org/10.1109/taai54685.2021.00029","url":null,"abstract":"This paper presents the realization of artificial intelligence (AI) in the Agricultural Industry driven by the chatbot impact on user experience. The study implemented past evaluation methods of Retrieval-based and Generative-based AI chatbots, including efficiency and construction costs. determine the extent of the customer satisfaction factors that successfully influence chatbot acceptance in Taiwan. The results of an empirical study demonstrated that an integrated method of retrieval and natural generation is the most suitable for agricultural services. In addition to the user's perception, the chatbot must consider the overall value evaluation, which Hybrid retrieval-generation can also achieve the most appropriate system efficiency and user satisfaction.","PeriodicalId":343821,"journal":{"name":"2021 International Conference on Technologies and Applications of Artificial Intelligence (TAAI)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132215474","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：