Pub Date : 2022-11-05DOI: 10.1109/iSAI-NLP56921.2022.9960262
R. Siricharoenchai, Panchapawn Chatsuwan, Paramet Tanwanont, Sarunruk Janbradab, Navaporn Surasvadi, S. Thajchayapong
In this work, a system is created to suggest product/ service code and industrial classification code for Thai language. The system can suggest UNSPSC and TSIC codes relevant to query terms via indexing search. Techniques used in this work are based on knowledge of text processing and text similarity, as well as indexing. Through a complexity analysis, the system has been proved efficient as it can retrieve data about 1,000 times faster than traditional methods. Furthermore, Mean Reciprocal Rank (MRR) was employed to evaluate the search results of 1,000 products and services. The results showed that the proposed system achieved the MRR of 0.46, indicating the relevant search result is approximately in the second or third rank. Currently, the proposed system has been implemented as a part of SMEs registration process in the OSMEP website to support Thai SMEs to access government procurement.
{"title":"Product and Industrial Classification Code Suggestion System for Thai Language","authors":"R. Siricharoenchai, Panchapawn Chatsuwan, Paramet Tanwanont, Sarunruk Janbradab, Navaporn Surasvadi, S. Thajchayapong","doi":"10.1109/iSAI-NLP56921.2022.9960262","DOIUrl":"https://doi.org/10.1109/iSAI-NLP56921.2022.9960262","url":null,"abstract":"In this work, a system is created to suggest product/ service code and industrial classification code for Thai language. The system can suggest UNSPSC and TSIC codes relevant to query terms via indexing search. Techniques used in this work are based on knowledge of text processing and text similarity, as well as indexing. Through a complexity analysis, the system has been proved efficient as it can retrieve data about 1,000 times faster than traditional methods. Furthermore, Mean Reciprocal Rank (MRR) was employed to evaluate the search results of 1,000 products and services. The results showed that the proposed system achieved the MRR of 0.46, indicating the relevant search result is approximately in the second or third rank. Currently, the proposed system has been implemented as a part of SMEs registration process in the OSMEP website to support Thai SMEs to access government procurement.","PeriodicalId":399019,"journal":{"name":"2022 17th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132861690","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 : 2022-11-05DOI: 10.1109/iSAI-NLP56921.2022.9960276
J. Gatedee, Kanokwan Jaiping, Sumana Kasemsawasdi, Aungsana Yothinarak, J. Netsawang, Supanit Angsirikul, Rachasak Somyanonthanakul
Dyslipidemia leads to cardiovascular disease with several complications which include sudden cardiac death, acute myocardial infarction, and strokes. The primary evaluation tool for dyslipidemia is a fasting lipid panel which consists of total cholesterol (TC), (LDL-C), (HDL-C), and triglycerides (TG). However, the relationship between a fasting lipid panel and elevated uric acid has not been comprehensively investigated. This work investigates the relationship between serum uric acid (SUA) and a fasting lipid panel in the elderly patients in Thailand. A rule-based machine learning technique called association rule mining was used to define patterns in the rules discovered. The results showed a significant positive relationship for SUA with TG, TC and LDL levels, and an inverse relationship for SUA with HDL. Early prevention of hyperuricemia and dyslipidemia may be helpful to reduce the incidence of associated cardiovascular diseases.
{"title":"Association of Serum Uric Acid and Lipid Parameters in Patients at Lamphun Hospital, Thailand","authors":"J. Gatedee, Kanokwan Jaiping, Sumana Kasemsawasdi, Aungsana Yothinarak, J. Netsawang, Supanit Angsirikul, Rachasak Somyanonthanakul","doi":"10.1109/iSAI-NLP56921.2022.9960276","DOIUrl":"https://doi.org/10.1109/iSAI-NLP56921.2022.9960276","url":null,"abstract":"Dyslipidemia leads to cardiovascular disease with several complications which include sudden cardiac death, acute myocardial infarction, and strokes. The primary evaluation tool for dyslipidemia is a fasting lipid panel which consists of total cholesterol (TC), (LDL-C), (HDL-C), and triglycerides (TG). However, the relationship between a fasting lipid panel and elevated uric acid has not been comprehensively investigated. This work investigates the relationship between serum uric acid (SUA) and a fasting lipid panel in the elderly patients in Thailand. A rule-based machine learning technique called association rule mining was used to define patterns in the rules discovered. The results showed a significant positive relationship for SUA with TG, TC and LDL levels, and an inverse relationship for SUA with HDL. Early prevention of hyperuricemia and dyslipidemia may be helpful to reduce the incidence of associated cardiovascular diseases.","PeriodicalId":399019,"journal":{"name":"2022 17th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125797335","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 : 2022-11-05DOI: 10.1109/iSAI-NLP56921.2022.9960257
S. Deepaisarn, Paphana Yiwsiw, Chanon Tantiwattanapaibul, Suphachok Buaruk, Virach Sornlertlamvanich
As a recent trend in urbanization and intelligent technologies, smart lighting systems have been implemented in many major cities to support smart urban environments. This research developed a web application platform for data visualization and lighting device monitoring at Thammasat Uni-versity, Rangsit Campus, Thailand. This implementation provides administrative and operative staff with an all-in-one platform through a convenient interface for monitoring, controlling, and collecting data from area devices and sensors. Platform devel-opment was divided into two sections: back-end application, providing application programming interface (API) endpoints, and front-end application, offering an interface for interacting with on-campus staff. Finally, the web application was deployed on a cloud platform so that responsible persons may access it on any device and acquire data in real time. Given the platform's capabilities, further data analytics may be proposed for building a smarter lighting system.
{"title":"Smart Street Light Monitoring and Visualization Platform for Campus Management","authors":"S. Deepaisarn, Paphana Yiwsiw, Chanon Tantiwattanapaibul, Suphachok Buaruk, Virach Sornlertlamvanich","doi":"10.1109/iSAI-NLP56921.2022.9960257","DOIUrl":"https://doi.org/10.1109/iSAI-NLP56921.2022.9960257","url":null,"abstract":"As a recent trend in urbanization and intelligent technologies, smart lighting systems have been implemented in many major cities to support smart urban environments. This research developed a web application platform for data visualization and lighting device monitoring at Thammasat Uni-versity, Rangsit Campus, Thailand. This implementation provides administrative and operative staff with an all-in-one platform through a convenient interface for monitoring, controlling, and collecting data from area devices and sensors. Platform devel-opment was divided into two sections: back-end application, providing application programming interface (API) endpoints, and front-end application, offering an interface for interacting with on-campus staff. Finally, the web application was deployed on a cloud platform so that responsible persons may access it on any device and acquire data in real time. Given the platform's capabilities, further data analytics may be proposed for building a smarter lighting system.","PeriodicalId":399019,"journal":{"name":"2022 17th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP)","volume":"393 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124385337","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 : 2022-11-05DOI: 10.1109/iSAI-NLP56921.2022.9960246
Teetouch Jaknamon, S. Marukatat
For problems with image captioning is a technique that has been used for a long time. In the past, there was a way to use convolutional neural network (CNN) for feature extraction and recurrent neural network (RNN) for generating text, and especially in Thai language, It has to be developed further in the era of the popular use of transformers. This paper proposes an end-to-end image captioning with pretrained vision Transformers (ViT) and text transformers in Thai language models namely ThaiTC, Which leverages the transformer architecture both. We has experiment pretrained vision transformer and text transformer in Thai language that best for Thai image captioning and tested on 3 Thai image captioning datasets 1) Travel 2) Food 3) Flickr 30k(t$r$ anslate) with different challenges. Includes freeze vision transformers weight training for image captioning dataset training with less image features, From the experiment, We found that ThaiTC performed much better in the Food and Flickr30k datasets than the Travel datasets, Which allowed us to automatically create subtitles about food and travel.
{"title":"ThaiTC:Thai Transformer-based Image Captioning","authors":"Teetouch Jaknamon, S. Marukatat","doi":"10.1109/iSAI-NLP56921.2022.9960246","DOIUrl":"https://doi.org/10.1109/iSAI-NLP56921.2022.9960246","url":null,"abstract":"For problems with image captioning is a technique that has been used for a long time. In the past, there was a way to use convolutional neural network (CNN) for feature extraction and recurrent neural network (RNN) for generating text, and especially in Thai language, It has to be developed further in the era of the popular use of transformers. This paper proposes an end-to-end image captioning with pretrained vision Transformers (ViT) and text transformers in Thai language models namely ThaiTC, Which leverages the transformer architecture both. We has experiment pretrained vision transformer and text transformer in Thai language that best for Thai image captioning and tested on 3 Thai image captioning datasets 1) Travel 2) Food 3) Flickr 30k(t$r$ anslate) with different challenges. Includes freeze vision transformers weight training for image captioning dataset training with less image features, From the experiment, We found that ThaiTC performed much better in the Food and Flickr30k datasets than the Travel datasets, Which allowed us to automatically create subtitles about food and travel.","PeriodicalId":399019,"journal":{"name":"2022 17th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP)","volume":"46 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121220917","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 : 2022-11-05DOI: 10.1109/iSAI-NLP56921.2022.9960247
Wirit Khongcharoen, Chanatip Saetia, Tawunrat Chalothorn, P. Buabthong
Question Answering over Knowledge Graphs (KGQA) extracts the answer entity directly from the graph, given a natural language question, offering scalability to applications that need to readily provide information to the end users, such as chatbots. Nevertheless, KGQA specifically designed for Knowledge Graphs in Thai has not yet been well investigated. In this paper, we adapt multi-hop KGQA using Graph Embedding to handle Thai dataset while being able to extract answer entities that do not have explicit relation to the head node. We also construct a Thai Knowledge Graph with the ontology based on retail banking products. The model achieves a HITS @ 1 score of 80.8 on our annotated dataset. The results confirm that, aside from reaching multi-hop answers, using Graph Embedding in KGQA helps improve the overall score, especially in sparse Knowledge Graphs. Moreover, augmenting the training questions to include more entities in the graph can further help increase the performance.
{"title":"Question Answering over Knowledge Graphs for Thai Retail Banking Products","authors":"Wirit Khongcharoen, Chanatip Saetia, Tawunrat Chalothorn, P. Buabthong","doi":"10.1109/iSAI-NLP56921.2022.9960247","DOIUrl":"https://doi.org/10.1109/iSAI-NLP56921.2022.9960247","url":null,"abstract":"Question Answering over Knowledge Graphs (KGQA) extracts the answer entity directly from the graph, given a natural language question, offering scalability to applications that need to readily provide information to the end users, such as chatbots. Nevertheless, KGQA specifically designed for Knowledge Graphs in Thai has not yet been well investigated. In this paper, we adapt multi-hop KGQA using Graph Embedding to handle Thai dataset while being able to extract answer entities that do not have explicit relation to the head node. We also construct a Thai Knowledge Graph with the ontology based on retail banking products. The model achieves a HITS @ 1 score of 80.8 on our annotated dataset. The results confirm that, aside from reaching multi-hop answers, using Graph Embedding in KGQA helps improve the overall score, especially in sparse Knowledge Graphs. Moreover, augmenting the training questions to include more entities in the graph can further help increase the performance.","PeriodicalId":399019,"journal":{"name":"2022 17th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP)","volume":"55 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121937440","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 : 2022-11-05DOI: 10.1109/isai-nlp56921.2022.9960269
{"title":"Welcome Message from the ISAI-NLP 2022 General and Conference Chairs","authors":"","doi":"10.1109/isai-nlp56921.2022.9960269","DOIUrl":"https://doi.org/10.1109/isai-nlp56921.2022.9960269","url":null,"abstract":"","PeriodicalId":399019,"journal":{"name":"2022 17th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134471212","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}
In this paper, we present a machine intelligent system that can automatically classify sugarcane images into predefined categories. This system is developed in order to facilitate the operation in sugar manufacturing factories and can be beneficial to the sugar industry as a whole. The software system consists of the core computer vision module and other compounds, such as user interfaces and database management. To develop the core module, we apply deep learning models based on convolutional neural networks, which are currently state-of-the-art models for computer vision. The best models trained and evaluated on our sugarcane datasets achieve more than 90% multi-class accuracy in almost all settings. We have incorporated the trained model into the prototype system and successfully installed the system to test operating at one of the major sugar manufacturing facilities in the previous sugarcane harvesting season.
{"title":"Sugarcane Classification for On-Site Assessment Using Computer Vision","authors":"Piyapoj Kasempakdeepong, Pondsulee Ponchaiyapruek, Pattamon Viriyothai, Anuwat Songchumrong, Pittipol Kantavat, Prasertsak Pungprasertying","doi":"10.1109/iSAI-NLP56921.2022.9960252","DOIUrl":"https://doi.org/10.1109/iSAI-NLP56921.2022.9960252","url":null,"abstract":"In this paper, we present a machine intelligent system that can automatically classify sugarcane images into predefined categories. This system is developed in order to facilitate the operation in sugar manufacturing factories and can be beneficial to the sugar industry as a whole. The software system consists of the core computer vision module and other compounds, such as user interfaces and database management. To develop the core module, we apply deep learning models based on convolutional neural networks, which are currently state-of-the-art models for computer vision. The best models trained and evaluated on our sugarcane datasets achieve more than 90% multi-class accuracy in almost all settings. We have incorporated the trained model into the prototype system and successfully installed the system to test operating at one of the major sugar manufacturing facilities in the previous sugarcane harvesting season.","PeriodicalId":399019,"journal":{"name":"2022 17th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133230488","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 : 2022-11-05DOI: 10.1109/iSAI-NLP56921.2022.9960286
Nalin Srun, Sotheara Leang, Ye Kyaw Thu, Sethserey Sam
Convolutional Neural Networks have been proven to successfully capture spatial aspects of the speech signal and eliminate spectral variations across speakers for Automatic Speech Recognition. In this study, we investigate the Convolutional Neural Net-work with Time Delay Neural Network for an acoustic model to deal with large vocabulary continuous speech recognition for Khmer. Our idea is to use Convolutional Neural Networks to extract local features of the speech signal, whereas Time Delay Neural Networks capture long temporal correlations between acoustic events. The experimental results show that the suggested net-work outperforms the Time Delay Neural Network and achieves an average relative improvement of 14% across test sets.
{"title":"Convolutional Time Delay Neural Network for Khmer Automatic Speech Recognition","authors":"Nalin Srun, Sotheara Leang, Ye Kyaw Thu, Sethserey Sam","doi":"10.1109/iSAI-NLP56921.2022.9960286","DOIUrl":"https://doi.org/10.1109/iSAI-NLP56921.2022.9960286","url":null,"abstract":"Convolutional Neural Networks have been proven to successfully capture spatial aspects of the speech signal and eliminate spectral variations across speakers for Automatic Speech Recognition. In this study, we investigate the Convolutional Neural Net-work with Time Delay Neural Network for an acoustic model to deal with large vocabulary continuous speech recognition for Khmer. Our idea is to use Convolutional Neural Networks to extract local features of the speech signal, whereas Time Delay Neural Networks capture long temporal correlations between acoustic events. The experimental results show that the suggested net-work outperforms the Time Delay Neural Network and achieves an average relative improvement of 14% across test sets.","PeriodicalId":399019,"journal":{"name":"2022 17th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP)","volume":"119 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116378759","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 : 2022-11-05DOI: 10.1109/iSAI-NLP56921.2022.9960256
Kasidit Eiam-o-pas, Nuchjarin Intalar, C. Jeenanunta
This paper examines the main factors affecting dental-related users” acceptance of dental appointment technology as a means for receiving dental appointment services. A questionnaire was developed based on the Technology Acceptance Model (TAM) and incorporated perceived useful features to understand user characteristics, acceptance, and usage behavior of a dental appointment application. A proposed research model and hypotheses were tested with a sample of 555 customers of a dental clinic in Bangkok and Metropolitan area using descriptive analysis, factor analysis, and multiple regression. The findings show that perceived ease of use, perceived usefulness and perceived value have significant effects on the acceptance of a dental appointment application. However, the application feature has no direct effect on the intention to use. Results can be used as a reference to develop dental appointment services that align with the needs of the users.
{"title":"Factors Affecting Acceptance of Dental Appointment Application among Users in Bangkok and Metropolitan Area","authors":"Kasidit Eiam-o-pas, Nuchjarin Intalar, C. Jeenanunta","doi":"10.1109/iSAI-NLP56921.2022.9960256","DOIUrl":"https://doi.org/10.1109/iSAI-NLP56921.2022.9960256","url":null,"abstract":"This paper examines the main factors affecting dental-related users” acceptance of dental appointment technology as a means for receiving dental appointment services. A questionnaire was developed based on the Technology Acceptance Model (TAM) and incorporated perceived useful features to understand user characteristics, acceptance, and usage behavior of a dental appointment application. A proposed research model and hypotheses were tested with a sample of 555 customers of a dental clinic in Bangkok and Metropolitan area using descriptive analysis, factor analysis, and multiple regression. The findings show that perceived ease of use, perceived usefulness and perceived value have significant effects on the acceptance of a dental appointment application. However, the application feature has no direct effect on the intention to use. Results can be used as a reference to develop dental appointment services that align with the needs of the users.","PeriodicalId":399019,"journal":{"name":"2022 17th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP)","volume":"23 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123272206","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 : 2022-11-05DOI: 10.1109/iSAI-NLP56921.2022.9960275
Mengmeng Gong, Hui Song, Haoran Zhou, Bo Xu
VAE (Variational Autoencoder) and CVAE (Conditional V AE) encode the sentence with the latent variable to generate response in Dialogue. However, studies have shown that the latent variables obtained are more inclined to remember the first words and the length of the sentence, and only represents limited local features. In order to alleviate this problem, we propose to involve contrastive learning to generate positive and negative samples for training process, which enriches the latent variables representation with the global information of sentence and generates more relevant response. On the other hand, those generative models do not consider emotional information of dialogue, a sentiment discrimination module is introduced in our model to maintain the emotional consistency. Experiments on two public datasets - DailyDialog and PERSONA-CHAT demonstrate the effectiveness of our method, the evaluation results of BLEU and Rouge are both improved. The sentiment discrimination network also forces the model to generating emotional consistency response with share embedding.
{"title":"Enhancing Response Relevance and Emotional Consistency for Dialogue Response Generation","authors":"Mengmeng Gong, Hui Song, Haoran Zhou, Bo Xu","doi":"10.1109/iSAI-NLP56921.2022.9960275","DOIUrl":"https://doi.org/10.1109/iSAI-NLP56921.2022.9960275","url":null,"abstract":"VAE (Variational Autoencoder) and CVAE (Conditional V AE) encode the sentence with the latent variable to generate response in Dialogue. However, studies have shown that the latent variables obtained are more inclined to remember the first words and the length of the sentence, and only represents limited local features. In order to alleviate this problem, we propose to involve contrastive learning to generate positive and negative samples for training process, which enriches the latent variables representation with the global information of sentence and generates more relevant response. On the other hand, those generative models do not consider emotional information of dialogue, a sentiment discrimination module is introduced in our model to maintain the emotional consistency. Experiments on two public datasets - DailyDialog and PERSONA-CHAT demonstrate the effectiveness of our method, the evaluation results of BLEU and Rouge are both improved. The sentiment discrimination network also forces the model to generating emotional consistency response with share embedding.","PeriodicalId":399019,"journal":{"name":"2022 17th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125075886","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：