Few-Shot Dialogue Generation Without Annotated Data: A Transfer Learning Approach. Shalyminov, I., Lemon, O., Eshghi, A., & Lee, S. In Proceedings of SIGdial 2019, 2019. ACL. ![Few-Shot Dialogue Generation Without Annotated Data: A Transfer Learning Approach [link]](https://bibbase.org/img/filetypes/link.svg "link")
Website abstract bibtex Learning with minimal data is one of the key challenges in the development of practical, production-ready goal-oriented dialogue systems. In a real-world enterprise setting where dialogue systems are developed rapidly and are expected to work robustly for an ever-growing variety of domains, products, and scenarios, efficient learning from a limited number of examples becomes indispensable. In this paper, we introduce a technique to achieve state-of-the-art dialogue generation performance in a few-shot setup, without using any annotated data. We do this by leveraging background knowledge from a larger, more highly represented dialogue source --- namely, the MetaLWOz dataset. We evaluate our model on the Stanford Multi-Domain Dialogue Dataset, consisting of human-human goal-oriented dialogues in in-car navigation, appointment scheduling, and weather information domains. We show that our few-shot approach achieves state-of-the art results on that dataset by consistently outperforming the previous best model in terms of BLEU and Entity F1 scores, while being more data-efficient by not requiring any data annotation.
@inproceedings{
title = {Few-Shot Dialogue Generation Without Annotated Data: A Transfer Learning Approach},
type = {inproceedings},
year = {2019},
websites = {https://arxiv.org/abs/1908.05854},
publisher = {ACL},
city = {Stockholm, Sweden},
id = {99f33106-6acf-34d0-9a9a-371aa2cd305b},
created = {2019-08-16T08:41:19.815Z},
file_attached = {false},
profile_id = {d7d2e6da-aa5b-3ab3-b3f2-a5350adf574a},
last_modified = {2019-08-20T09:50:49.213Z},
read = {false},
starred = {false},
authored = {true},
confirmed = {true},
hidden = {false},
citation_key = {Shalyminov.etal19b},
private_publication = {false},
abstract = {Learning with minimal data is one of the key challenges in the development of practical, production-ready goal-oriented dialogue systems. In a real-world enterprise setting where dialogue systems are developed rapidly and are expected to work robustly for an ever-growing variety of domains, products, and scenarios, efficient learning from a limited number of examples becomes indispensable. In this paper, we introduce a technique to achieve state-of-the-art dialogue generation performance in a few-shot setup, without using any annotated data. We do this by leveraging background knowledge from a larger, more highly represented dialogue source --- namely, the MetaLWOz dataset. We evaluate our model on the Stanford Multi-Domain Dialogue Dataset, consisting of human-human goal-oriented dialogues in in-car navigation, appointment scheduling, and weather information domains. We show that our few-shot approach achieves state-of-the art results on that dataset by consistently outperforming the previous best model in terms of BLEU and Entity F1 scores, while being more data-efficient by not requiring any data annotation.},
bibtype = {inproceedings},
author = {Shalyminov, Igor and Lemon, Oliver and Eshghi, Arash and Lee, Sungjin},
booktitle = {Proceedings of SIGdial 2019},
keywords = {Dialogue System,Few-Shot,Generation,Knowledge Transfer,Transfer Learning}
}
Downloads: 0
{"_id":"opm2nz97aiBkamAdB","bibbaseid":"shalyminov-lemon-eshghi-lee-fewshotdialoguegenerationwithoutannotateddataatransferlearningapproach-2019","author_short":["Shalyminov, I.","Lemon, O.","Eshghi, A.","Lee, S."],"bibdata":{"title":"Few-Shot Dialogue Generation Without Annotated Data: A Transfer Learning Approach","type":"inproceedings","year":"2019","websites":"https://arxiv.org/abs/1908.05854","publisher":"ACL","city":"Stockholm, Sweden","id":"99f33106-6acf-34d0-9a9a-371aa2cd305b","created":"2019-08-16T08:41:19.815Z","file_attached":false,"profile_id":"d7d2e6da-aa5b-3ab3-b3f2-a5350adf574a","last_modified":"2019-08-20T09:50:49.213Z","read":false,"starred":false,"authored":"true","confirmed":"true","hidden":false,"citation_key":"Shalyminov.etal19b","private_publication":false,"abstract":"Learning with minimal data is one of the key challenges in the development of practical, production-ready goal-oriented dialogue systems. In a real-world enterprise setting where dialogue systems are developed rapidly and are expected to work robustly for an ever-growing variety of domains, products, and scenarios, efficient learning from a limited number of examples becomes indispensable. In this paper, we introduce a technique to achieve state-of-the-art dialogue generation performance in a few-shot setup, without using any annotated data. We do this by leveraging background knowledge from a larger, more highly represented dialogue source --- namely, the MetaLWOz dataset. We evaluate our model on the Stanford Multi-Domain Dialogue Dataset, consisting of human-human goal-oriented dialogues in in-car navigation, appointment scheduling, and weather information domains. We show that our few-shot approach achieves state-of-the art results on that dataset by consistently outperforming the previous best model in terms of BLEU and Entity F1 scores, while being more data-efficient by not requiring any data annotation.","bibtype":"inproceedings","author":"Shalyminov, Igor and Lemon, Oliver and Eshghi, Arash and Lee, Sungjin","booktitle":"Proceedings of SIGdial 2019","keywords":"Dialogue System,Few-Shot,Generation,Knowledge Transfer,Transfer Learning","bibtex":"@inproceedings{\n title = {Few-Shot Dialogue Generation Without Annotated Data: A Transfer Learning Approach},\n type = {inproceedings},\n year = {2019},\n websites = {https://arxiv.org/abs/1908.05854},\n publisher = {ACL},\n city = {Stockholm, Sweden},\n id = {99f33106-6acf-34d0-9a9a-371aa2cd305b},\n created = {2019-08-16T08:41:19.815Z},\n file_attached = {false},\n profile_id = {d7d2e6da-aa5b-3ab3-b3f2-a5350adf574a},\n last_modified = {2019-08-20T09:50:49.213Z},\n read = {false},\n starred = {false},\n authored = {true},\n confirmed = {true},\n hidden = {false},\n citation_key = {Shalyminov.etal19b},\n private_publication = {false},\n abstract = {Learning with minimal data is one of the key challenges in the development of practical, production-ready goal-oriented dialogue systems. In a real-world enterprise setting where dialogue systems are developed rapidly and are expected to work robustly for an ever-growing variety of domains, products, and scenarios, efficient learning from a limited number of examples becomes indispensable. In this paper, we introduce a technique to achieve state-of-the-art dialogue generation performance in a few-shot setup, without using any annotated data. We do this by leveraging background knowledge from a larger, more highly represented dialogue source --- namely, the MetaLWOz dataset. We evaluate our model on the Stanford Multi-Domain Dialogue Dataset, consisting of human-human goal-oriented dialogues in in-car navigation, appointment scheduling, and weather information domains. We show that our few-shot approach achieves state-of-the art results on that dataset by consistently outperforming the previous best model in terms of BLEU and Entity F1 scores, while being more data-efficient by not requiring any data annotation.},\n bibtype = {inproceedings},\n author = {Shalyminov, Igor and Lemon, Oliver and Eshghi, Arash and Lee, Sungjin},\n booktitle = {Proceedings of SIGdial 2019},\n keywords = {Dialogue System,Few-Shot,Generation,Knowledge Transfer,Transfer Learning}\n}","author_short":["Shalyminov, I.","Lemon, O.","Eshghi, A.","Lee, S."],"urls":{"Website":"https://arxiv.org/abs/1908.05854"},"biburl":"https://bibbase.org/service/mendeley/d7d2e6da-aa5b-3ab3-b3f2-a5350adf574a","bibbaseid":"shalyminov-lemon-eshghi-lee-fewshotdialoguegenerationwithoutannotateddataatransferlearningapproach-2019","role":"author","keyword":["Dialogue System","Few-Shot","Generation","Knowledge Transfer","Transfer Learning"],"metadata":{"authorlinks":{}}},"bibtype":"inproceedings","biburl":"https://bibbase.org/service/mendeley/d7d2e6da-aa5b-3ab3-b3f2-a5350adf574a","dataSources":["ya2CyA73rpZseyrZ8","TcQYToyGTfqDApS68","gQ3XnmcCc7p5JYvBa","pB6WyiWKHxyFpemKA","2252seNhipfTmjEBQ"],"keywords":["dialogue system","few-shot","generation","knowledge transfer","transfer learning"],"search_terms":["few","shot","dialogue","generation","without","annotated","data","transfer","learning","approach","shalyminov","lemon","eshghi","lee"],"title":"Few-Shot Dialogue Generation Without Annotated Data: A Transfer Learning Approach","year":2019}