近年來，基於類神經網路方法在任務導向的對話系統中顯示出了卓越的成功，產生任務導向的對話在很大程度上依賴於訪問外部知識庫來檢索與任務相關的信息。然而當開發現實世界的任務導向的對話系統時，通常涉及訪問大型外部知識庫，而這些大型知識庫不能簡單地通過諸如存儲記憶網絡機制之類的類神經網路方法進行編碼。為了緩解上述的問題，在本文中我們提出一個端到端訓練的文本轉換結構化查詢語言引導框架，用以訓練類神經任務導向對話系統能夠用生成的結構化查詢語言與知識庫互動以便獲得資料。具體來說，類神經任務導向對話系統首先學習詢問並確認客戶的意圖，然後動態決定何時將用客戶的需求限制轉換成可執行的結構化查詢語言，藉此從知識庫中獲取相關信息。借助我們的方法，類神經任務導向對話系統不需要將全部知識庫整合進系統，而可以只用少量並更準確的查詢資料結果，有效率的產生有用的對話回覆。我們在 AirDialogue 資料集上評估所的提出的方法，該資料集是 Google 釋出的一個大型任務導向語料庫，其中包含客戶與系統代理預訂機票的對話。實驗表明我們提出的方法在任務準確性和 BLEU 得分方面比之前的模型方法有顯著提高，這不僅顯示我們提出的方法生成的對話有很好的質量，還展示了完成給定任務的能力。

Despite recent success in neural task-oriented dialogue systems, developing such a real-world system involves accessing large-scale knowledge bases (KBs), which cannot be simply encoded by neural approaches, such as memory network mechanisms. To alleviate the above problem, we propose \airc, an end-to-end trainable text-to-SQL guided framework to learn a neural agent that interacts with KBs using the generated SQL queries. Specifically, the neural agent first learns to ask and confirm the customer's intent during the multi-turn interactions, then dynamically determining when to ground the user constraints into executable SQL queries so as to fetch relevant information from KBs. With the help of our method, the agent can use less but more accurate fetched results to generate useful responses efficiently, instead of incorporating the entire KBs. We evaluate the proposed method on the AirDialogue dataset, a large corpus released by Google, containing the conversations of customers booking flight tickets from the agent. The experimental results show that \airc\ significantly improves over previous work in terms of accuracy and the BLEU score, which demonstrates not only the ability to achieve the given task but also the good quality of the generated dialogues.

1 Introduction ---- 1
2 Related Work ---- 5
2.1 Task-oriented Dialogue System ---- 5
2.2 Semantic Parsing in SQL ---- 9
3 The Proposed Framework ---- 10
3.1 System Architecture of AirConcierge ---- 10
3.2 Dialogue Encoder ---- 12
3.3 Dialogue State Tracker (Information Gate Module) ---- 13
3.4 SQL Generator ---- 14
3.5 Knowledge Base Memory Encoder ---- 15
3.6 Dialogue Decoder ---- 17
3.7 Dialogue Goal Generator ---- 17
3.8 Objective Function ---- 18
4 Experiments ---- 19
4.1 Dataset ---- 19
4.2 Training Details ---- 20
4.3 Evaluation ---- 21
4.4 Experimental Results: Accuracy ---- 23
4.5 Experimental Results: Scalability ---- 25
4.6 Supplementary ---- 27
5 Conclusions ---- 34
References ---- 35

[1] A. Bordes, Y.-L. Boureau, and W. Jason. Learning end-to-end goal-oriented dialog.
In ICLR, 2017.
[2] J. Chung, C¸ aglar G¨ulc¸ehre, K. Cho, and Y. Bengio. Empirical evaluation of gated
recurrent neural networks on sequence modeling. ArXiv, abs/1412.3555, 2014.
[3] R. Das, M. Zaheer, S. Reddy, and A. McCallum. Question answering on knowledge
bases and text using universal schema and memory networks. In ACL, 2017.
[4] A. Deoras and R. Sarikaya. Deep belief network based semantic taggers for spoken
language understanding. In INTERSPEECH, 2013.
[5] B. Dhingra, L. Li, X. Li, J. Gao, Y.-N. Chen, F. Ahmed, and L. Deng. Towards
end-to-end reinforcement learning of dialogue agents for information access. In ACL,
2017.
[6] E. Dinan, S. Roller, K. Shuster, A. Fan, M. Auli, and J. Weston. Wizard of wikipedia:
Knowledge-powered conversational agents. In ICLR, 2019.
[7] J. Dodge, A. Gane, X. Zhang, A. Bordes, S. Chopra, A. H. Miller, A. Szlam, and
J. Weston. Evaluating prerequisite qualities for learning end-to-end dialog systems.
CoRR, abs/1511.06931, 2016.
[8] M. Eric and C. D. Manning. Key-value retrieval networks for task-oriented dialogue.
In SIGDIAL, 2017.
[9] M. Ghazvininejad, C. Brockett, M.-W. Chang, B. Dolan, J. Gao, W.-t. Yih, and
M. Galley. A knowledge-grounded neural conversation model. In AAAI, 2018.
[10] W. Hwang, J. Yim, S. Park, and M. Seo. A comprehensive exploration on wikisql
with table-aware word contextualization. arXiv preprint arXiv:1902.01069, 2019.
[11] K. Kim, C. Lee, S. Jung, and G. G. Lee. A frame-based probabilistic framework for
spoken dialog management using dialog examples. In SIGDIAL Workshop, 2008.
[12] D. P. Kingma and J. Ba. Adam: A method for stochastic optimization. CoRR,
abs/1412.6980, 2015.
[13] W. Lei, X. Jin, Z. Ren, X. He, M.-Y. Kan, and D. Yin. Sequicity: Simplifying
task-oriented dialogue systems with single sequence-to-sequence architectures. In
ACL, 2018.
[14] X. Li, Y.-N. Chen, L. Li, J. Gao, and A. C¸ elikyilmaz. End-to-end task-completion
neural dialogue systems. ArXiv, abs/1703.01008, 2017.
[15] B. Liu and I. Lane. An end-to-end trainable neural network model with belief tracking
for task-oriented dialog. ArXiv, abs/1708.05956, 2017.
[16] A. Madotto, C.-S. Wu, and P. Fung. Mem2seq: Effectively incorporating knowledge
bases into end-to-end task-oriented dialog systems. ArXiv, abs/1804.08217, 2018.
[17] C. D. Manning and M. Eric. A copy-augmented sequence-to-sequence architecture
gives good performance on task-oriented dialogue. In EACL, 2017.
[18] B. McCann, N. S. Keskar, C. Xiong, and R. Socher. The natural language decathlon:
Multitask learning as question answering. arXiv preprint arXiv:1806.08730, 2018.
[19] Y. Mo, W. Yin, K. S. Hasan, C. d. Santos, B. Xiang, and B. Zhou. Improved neural
relation detection for knowledge base question answering. In ACL, 2017.
[20] A. Paszke, S. Gross, S. Chintala, G. Chanan, E. Yang, Z. Devito, Z. Lin, A. Desmaison,
L. Antiga, and A. Lerer. Automatic differentiation in pytorch. In NIPS-W,
2017.
[21] L. Qin, M. Galley, C. Brockett, X. Liu, X. Gao, B. Dolan, Y. Choi, and J. Gao.
Conversing by reading: Contentful neural conversation with on-demand machine
reading. In ACL, 2019.
[22] A. I. Rudnicky, E. H. Thayer, P. C. Constantinides, C. Tchou, R. Shern, K. A. Lenzo,
W. Xu, and A. H. Oh. Creating natural dialogs in the carnegie mellon communicator
system. In EUROSPEECH, 1999.
[23] I. Serban, A. Sordoni, Y. Bengio, A. C. Courville, and J. Pineau. Building end-to-end
dialogue systems using generative hierarchical neural network models. In AAAI,
2016.
[24] S. Sukhbaatar, A. Szlam, J. Weston, and R. Fergus. End-to-end memory networks.
In NIPS, 2015.
[25] H. Sun, B. Dhingra, M. Zaheer, K. Mazaitis, R. Salakhutdinov, and W. Cohen. Open
domain question answering using early fusion of knowledge bases and text. In
EMNLP, 2018.
[26] W.Wei, Q. V. Le, A. M. Dai, and J. Li. Airdialogue: An environment for goal-oriented
dialogue research. In EMNLP, 2018.
[27] T.-H. Wen, D. V. L. M. Rojas-Barahona, M. Gasic, N. Mrksic, P. hao Su, S. Ultes,
and S. J. Young. A network-based end-to-end trainable task-oriented dialogue system.
In EACL, 2016.
[28] J. Weston, S. Chorpa, and A. Bordes. Memory networks. arXiv:1410.3916, 2014.
[29] C.-S. Wu, R. Socher, and C. Xiong. Global-to-local memory pointer networks for
task-oriented dialogue. ArXiv, abs/1901.04713, 2019.
[30] X. Xu, C. Liu, and D. Song. Sqlnet: Generating structured queries from natural
language without reinforcement learning. In ICLR, 2018.
[31] X. Yang, Y.-N. Chen, D. Z. Hakkani-T¨ur, P. Crook, X. Li, J. Gao, and L. Deng.
End-to-end joint learning of natural language understanding and dialogue manager.
2017 IEEE International Conference on Acoustics, Speech and Signal Processing
(ICASSP), pages 5690–5694, 2017.
[32] S. J. Young, M. Gasic, B. Thomson, and J. D. Williams. Pomdp-based statistical
spoken dialog systems: A review. Proceedings of the IEEE, 101:1160–1179, 2013.
[33] T. Yu, Z. Li, Z. Zhang, R. Zhang, and D. Radev. Typesql: Knowledge-based typeaware
neural text-to-sql generation. In NAACL, 2018.
[34] T. Yu, R. Zhang, H. Y. Er, S. Li, E. Xue, B. Pang, X. V. Lin, Y. C. Tan, T. Shi, Z. Li,
Y. Jiang, M. Yasunaga, S. Shim, T. Chen, A. R. Fabbri, Z. Li, L. Chen, Y. Zhang,
S. Dixit, V. Zhang, C. Xiong, R. Socher, W. S. Lasecki, and D. R. Radev. Cosql: A
conversational text-to-sql challenge towards cross-domain natural language interfaces
to databases. In EMNLP/IJCNLP, 2019.
[35] T. Yu, R. Zhang, K. Yang, M. Yasunaga, D. Wang, Z. Li, J. Ma, I. Li, Q. Yao,
S. Roman, Z. Zhang, and D. R. Radev. Spider: A large-scale human-labeled dataset
for complex and cross-domain semantic parsing and text-to-sql task. In EMNLP,
2018.
[36] T. Yu, R. Zhang, M. Yasunaga, Y. C. Tan, X. V. Lin, S. Li, H. Er, I. Li, B. Pang,
T. Chen, E. Ji, S. Dixit, D. N. Proctor, S. Shim, J. Kraft, V. Zhang, C. Xiong, R. Socher,
and D. R. Radev. Sparc: Cross-domain semantic parsing in context. In ACL, 2019.
[37] T. Zhao and M. Esk´enazi. Towards end-to-end learning for dialog state tracking and
management using deep reinforcement learning. In SIGDIAL Conference, 2016.
[38] V. Zhong, C. Xiong, and R. Socher. Seq2sql: Generating structured queries from
natural language using reinforcement learning. ArXiv, abs/1709.00103, 2017.
[39] V. Zue. Conversational interfaces: advances and challenges. Proceedings of the
IEEE, 88:1166–1180, 2000.
[40] V. Zue, S. Seneff, J. R. Glass, J. Polifroni, C. Pao, T. J. Hazen, and I. L. Hetherington.
Juplter: a telephone-based conversational interface for weather information. IEEE
Trans. Speech Audio Process., 8:85–96, 2000.