本論文提出一個具有長記憶的多人對話系統，以多人點餐及家庭對話為例，我們利用長記憶性將對話過程紀錄下來，當使用者提到有關過去點餐或家庭對話時，能捕捉過去的對話紀錄給系統分析處理，另外也針對多人對話系統做探討，使用多人語言理解及多人對話管理，解決多人對話中常遇到的問題，提升使用者的體驗。
本對話系統主要可分為五個部分:(1)ASR(2)NLU (3)DM (4)NLG (5)TTS，其中我們將重點放在多人語言理解(NLU)、多人對話管理(DM)。而NLU的部分，有使用到BERT-NER，可以抓出使用者語句中的日期時間等，以幫助多人語言理解部分更好的分析語句資訊 ; 多人語言理解部分包含了領域(Domain)與意圖(Intent)偵測，確定了使用者的意圖後，我們給各意圖設定所需的槽填充(Slot Filling)以擷取到點餐或家庭(備忘錄)所需的資訊，然而在多人對話中有許多狀況是需要過去的對話資訊，因此在多人對話管理中我們使用ISU(Information States Updates)，管理目前對話進行到哪個流程並偵測是否要從過去的對話中撈取資料，且ISU也管理了對話進行到哪個步驟並做相對應的回答。最後本系統MOS的準確度評分為4.45分，速度的評分為4.74，證明我們的系統能給使用者不錯的體驗。

We proposed a multi-party dialogue system with long memory and took multi-party food ordering and home conversations as examples. We use long-term memory to record the conversation process. When the user mentions food orders or family conversations, it can capture past conversation records for system analysis and processing. In addition, we also discussed the multi-party dialogue system and used multi-party language understanding and multi-party dialogue management to solve common problems in multi-party conversations.
Our dialogue system can be divided into five parts: (1) ASR (2) NLU (3) DM (4) NLG (5) TTS. We focus on multi-party language understanding (NLU) and multi-party dialogue management (DM). In the part of NLU, we use BERT-NER to capture the date and time in the user’s sentence. And we also use domain, intent detection. After we determine the conversation’s domain and intention, we set the slot filling for each intention to capture the information for food ordering or home conversation. However, there are many situations in multi-party conversations need past conversation information. We use ISU (Information States Updates) in multi-party dialogue management to manage the current process of the conversation. And ISU also manages which step the conversation progresses and makes the corresponding answers. Finally, the accuracy score of the MOS of this system is 4.45 points. The score of system speed is 4.74 points. This prove that our system can give users a good experience.

[1] J. Weizenbaum, "ELIZA—a computer program for the study of natural language communication between man and machine,"Communications of the ACM, vol. 9, no. 1, pp. 36-45, 1966.
[2] López-Cózar, Ramón, et al. "Review of spoken dialogue systems." Loquens 1.2 (2014): 012.
[3] Tulshan, Amrita S., and Sudhir Namdeorao Dhage. "Survey on Virtual Assistant: Google Assistant, Siri, Cortana, Alexa." International Symposium on Signal Processing and Intelligent Recognition Systems. Springer, Singapore, 2018.
[4] Banchs, Rafael E., and Haizhou Li. "IRIS: a chat-oriented dialogue system based on the vector space model." Proceedings of the ACL 2012 System Demonstrations. 2012.
[5] Yan, Zhao, et al. "Building task-oriented dialogue systems for online shopping." Thirty-First AAAI Conference on Artificial Intelligence. 2017.
[6] Nio, Lasguido, et al. "Utilizing human-to-human conversation examples for a multi domain chat-oriented dialog system." IEICE TRANSACTIONS on Information and Systems 97.6 (2014): 1497-1505.
[7] Allen, James. Natural language understanding. Pearson, 1995.
[8] Traum, David. "Issues in multiparty dialogues." Workshop on Agent Communication Languages. Springer, Berlin, Heidelberg, 2003.
[9] Strauß, Petra-Maria, and Wolfgang Minker. "Dialogue Management for a Multi-Party Spoken Dialogue System." Proactive Spoken Dialogue Interaction in Multi-Party Environments. Springer, Boston, MA, 2010. 73-114.
[10] Devlin, Jacob, et al. "Bert: Pre-training of deep bidirectional transformers for language understanding." arXiv preprint arXiv:1810.04805 (2018).
[11] Vaswani, Ashish, et al. "Attention is all you need." Advances in neural information processing systems. 2017.
[12] Sun, J. "Jieba'Chinese word segmentation tool.(2012)." (2012).
[13] Silva, Catarina, and Bernardete Ribeiro. "The importance of stop word removal on recall values in text categorization." Proceedings of the International Joint Conference on Neural Networks, 2003.. Vol. 3. IEEE, 2003.
[14] Huang, Zhiheng, Wei Xu, and Kai Yu. "Bidirectional LSTM-CRF models for sequence tagging." arXiv preprint arXiv:1508.01991 (2015).
[15] Brenes, David J., Daniel Gayo-Avello, and Kilian Pérez-González. "Survey and evaluation of query intent detection methods." Proceedings of the 2009 workshop on Web Search Click Data. 2009.
[16] Jose, K. J., and K. S. Lakshmi. "Joint Slot Filling And Intent Prediction for Natural Language Understanding in Frames Dataset." 2018 International Conference on Inventive Research in Computing Applications (ICIRCA). IEEE, 2018..
[17] Traum, David R., and Staffan Larsson. "The information state approach to dialogue management." Current and new directions in discourse and dialogue. Springer, Dordrecht, 2003. 325-353.
[18] Georgila, Kallirroi, James Henderson, and Oliver Lemon. "Learning user simulations for information state update dialogue systems." Ninth European Conference on Speech Communication and Technology. 2005.
[19] McDonald, David D. "Natural Language Generation." Handbook of Natural Language Processing 2 (2010): 121-144.
[20] Brill, Eric D., and Matthew R. Richardson. "Table querying." U.S. Patent No. 7,792,829. 7 Sep. 2010.