簡易檢索 / 詳目顯示

回結果列表
研究生： 鄭曜祥
Cheng, Yao-Hsiang
論文名稱： 基於SIP架構及友善對話介面之多線路多輪問答電話訪問機器人
Multi-line and Multi-turn Question Answering Telephone Interview Bot Based on SIP Architecture and Friendly Dialogue Interface
指導教授： 王駿發
Wang, Jhing-Fa
學位類別： 碩士
Master
系所名稱： 電機資訊學院 - 電機工程學系
Department of Electrical Engineering
論文出版年： 2021
畢業學年度： 109
語文別： 英文
論文頁數： 60
中文關鍵詞： 問卷調查 、電話訪問 、VoIP 、SIP 、多輪問答
外文關鍵詞： Questionnaire Survey, Telephone Interview, VoIP, SIP, Multi-turn Question Answering
相關次數： 點閱：247  下載：0 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報

    • 現今政府或企業經常利用問卷來調查民眾對於議題或是產品的看法，而電話訪問正是常用來做問卷調查的方式之一。本研究提出一個基於SIP架構及友善對話介面之多線路多輪問答電話訪問機器人，系統主要分為四個部分。第一部分是前處理，根據待進行訪問的問卷去事先生成在後續的對話上所需要的資料。第二部分是建構VoIP軟體電話，目的是能夠透過SIP架構去實現在軟體端多線路同時撥打電話的功能。此外，系統也透過友善對話介面串接前端的軟體電話模組與後端的對話模組來自動進行電話訪問，並帶給受訪者一些友善的體驗。第三部分是多輪問答對話，由對話模組主動根據問卷中的題目來詢問受訪者，再經由混合式回答匹配來判斷受訪者的回答所匹配到的題目選項，最後決策出下一輪要詢問的題目，達到多輪對話的效果。第四部份是通聯紀錄與報表生成，負責將電話訪問過程中的通話狀態及作答情形記錄下來。另外本系統在電話訪問結束後，可以根據電話訪問的結果進行多線路的資料合併、統計與查詢，讓我們可以快速了解電話訪問的情況。實驗結果顯示本系統於實驗環境下可以同時進行20線路的電話訪問。在多輪問答對話上，本系統在實驗中最高可達到88.89%的對話完成率。最後本系統在使用者滿意度測試(MOS)中平均得到4.49分，表明我們的系統可以在實際應用時帶給使用者良好的感受。

    Nowadays, governments or enterprises often use questionnaires to investigate people's views on issues or products, and telephone interview is one of the most commonly used methods for questionnaire surveys. This study proposes a multi-line and multi-turn question answering telephone interview bot based on SIP architecture and friendly dialogue interface. The system is mainly divided into four parts. The first part is pre-processing, according to the questionnaire to be interviewed to generate the data needed in the follow-up dialogue in advance. The second part is to build a VoIP soft phone, the purpose is to realize the function of making calls on the software side with multiple SIP lines at the same time through the SIP architecture. In addition, the system also connects the front-end softphone module with the back-end dialogue module through a friendly dialogue interface to automatically conduct the telephone interview, and brings a friendly experience to the respondents. The third part is multi-turn question answering dialogue. If the call is answered, the dialogue module will actively ask the respondent according to the questions in the questionnaire, and then through hybrid answer matching to determine the option matched by the respondent’s answer. Finally, the dialogue module decides the question to be asked in the next turn to achieve the effect of multi-turn dialogue. The fourth part is call detail record and table generation, which is responsible for recording the call status and answering situation during the telephone interview. In addition, the system performs post-processing according to the telephone interview results of all lines after the execution, so that we can quickly understand the situation of the telephone interview. The experimental results show that the system can reach to twenty lines at the same time for the telephone interview under the experimental environment. In multi-turn question answering dialogue, our system can achieve a maximum dialogue completion rate of 88.89% in the experiments. Finally, the average MOS score of our system is 4.49, indicating that our system can give users a good experience in practical applications.

    中文摘要 I Abstract III 誌謝 V Content VI Table List IX Figure List XI Chapter1 Introduction 1 1.1 Background 1 1.2 Motivation 2 1.3 Objectives 2 1.4 Organization 3 Chapter2 Related Work 4 2.1 Questionnaire and Telephone Interview 4 2.1.1 Questionnaire 4 2.1.2 Telephone Interview 6 2.2 VoIP and SIP 7 2.2.1 VoIP 7 2.2.2 SIP Messages 8 2.2.3 SIP Components 9 2.3 Answer Matching 12 Chapter3 Multi-line and Multi-turn Question Answering Telephone Interview Bot Based on SIP Architecture and Friendly Dialogue Interface 14 3.1 System Overview 14 3.1.1 Pre-processing Module 15 3.1.2 Multi-line SIP Softphone Module 15 3.1.3 Multi-turn Question Answering Dialogue Module 15 3.1.4 Call Detail Record and Table Generation Module 16 3.1.5 Post-processing Module 16 3.2 Pre-processing Module 16 3.2.1 Frame Overview 16 3.2.2 Audio Pre-Processing 17 3.2.3 Data Augmentation 18 3.2.4 State Transition Design 20 3.3 Multi-line SIP Softphone Module 21 3.3.1 Frame Overview 21 3.3.2 Telephone Interview Scheduling 22 3.3.3 SIP Outbound-Call Connection 23 3.3.4 Media Session 26 3.3.5 Friendly Dialogue Interface 27 3.3.6 Multi-line Execution 29 3.4 Multi-turn Question Answering Dialogue Module 30 3.4.1 Frame Overview 30 3.4.2 Disconnect Handler 31 3.4.3 Questionnaire Audio Player and Response Recorder 31 3.4.4 Noise Reduction and Speech Recognition 36 3.4.5 Hybrid Answer Matching 37 3.4.6 Next Turn Decision 44 3.5 Call Detail Record and Table Generation Module 45 3.5.1 Frame Overview 45 3.5.2 Full Phone Call Recorder 45 3.5.3 Call Detail Record 46 3.6 Post-processing Module 47 3.6.1 Frame Overview 47 3.6.2 Table Merging and Filtering 48 3.6.3 Analysis and Inquiry 48 Chapter4 Experimental Results 50 4.1 Experiment of CPU Usage for Multi-line Execution 50 4.2 Experiment of Answer Matching 51 4.2.1 Evaluation methods for answer matching 51 4.2.2 Experimental Results for answer matching 52 4.3 Experiment of Multi-turn Question Answering Dialogue 52 4.3.1 Evaluation methods for Multi-turn Question Answering Dialogue 52 4.3.2 Experimental Results for Multi-turn Question Answering Dialogue 53 4.4 Experiment of MOS for Proposed System 54 Chapter5 Conclusions 56 5.1 Summary and Contributions 56 5.2 Future Works 56 References 57

    [1] T. S. Kristensen, "A questionnaire is more than a questionnaire," Scandinavian journal of public health, vol. 38, no. 3_suppl, pp. 149-155, 2010.
    [2] E. S. Block and L. Erskine, "Interviewing by telephone: Specific considerations, opportunities, and challenges," International journal of qualitative methods, vol. 11, no. 4, pp. 428-445, 2012.
    [3] H. B. Chin et al., "Piloting a computer assisted telephone interview: the FUCHSIA Women’s Study," BMC women's health, vol. 14, no. 1, pp. 1-7, 2014.
    [4] S. Jalendry and S. Verma, "A detail review on voice over internet protocol (voip)," International Journal of Engineering Trends and Technology (IJETT), vol. 23, no. 4, pp. 161-166, 2015.
    [5] R. López-Cózar, Z. Callejas, D. Griol, and J. F. Quesada, "Review of spoken dialogue systems," Loquens, vol. 1, no. 2, p. 012, 2014.
    [6] Z. Wei et al., "Task-oriented dialogue system for automatic diagnosis," in Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), 2018, pp. 201-207.
    [7] T. C. Wu et al., "Prevalence of food allergy in T aiwan: a questionnaire‐based survey," Internal medicine journal, vol. 42, no. 12, pp. 1310-1315, 2012.
    [8] T. Nanseki, S. Takeuchi, and Y. Shinozaki, "Business development, ICT use, and personnel training in agricultural corporations: an analysis of a nationwide questionnaire survey," Agricultural Information Research, vol. 22, no. 3, pp. 159-173, 2013.
    [9] C. Kennedy et al., "An evaluation of the 2016 election polls in the United States," Public Opinion Quarterly, vol. 82, no. 1, pp. 1-33, 2018.
    [10] A. M. Kazi and W. Khalid, "Questionnaire designing and validation," Journal of the Pakistan Medical Association, vol. 62, no. 5, p. 514, 2012.
    [11] S. Roopa and M. Rani, "Questionnaire designing for a survey," Journal of Indian Orthodontic Society, vol. 46, no. 4_suppl1, pp. 273-277, 2012.
    [12] P. A. Stehr-Green, J. K. Stehr-Green, A. Nelson, L. Alexander, G. C. Mejia, and P. D. MacDonald, "Developing a questionnaire," FOCUS on Field Epidemiology, vol. 2, no. 2, pp. 1-6, 2003.
    [13] R. L. Sinkowitz-Cochran, "Survey design: To ask or not to ask? That is the question…," Clinical Infectious Diseases, vol. 56, no. 8, pp. 1159-1164, 2013.
    [14] S. Brar et al., "Perinatal care for South Asian immigrant women and women born in Canada: telephone survey of users," Journal of Obstetrics and Gynaecology Canada, vol. 31, no. 8, pp. 708-716, 2009.
    [15] 黃文仙. 問卷調查資料的蒐集方法. Available: http://web.nchu.edu.tw/pweb/users/whh/lesson/10275.pdf.
    [16] C. J. Cerrada, J. Weinberg, K. J. Sherman, and R. B. Saper, "Inter-method reliability of paper surveys and computer assisted telephone interviews in a randomized controlled trial of yoga for low back pain," BMC research notes, vol. 7, no. 1, pp. 1-6, 2014.
    [17] D. Bednall, K. M. Plocinski, and S. Adam, "Response rates in telephone surveys: managing contactability," in ANZMAC 2004: marketing accountabilities and responsibilities, Wellington, 29 November-1 December 2004: proceedings, 2004: ANZMAC, pp. 1-7.
    [18] R. Corkrey and L. Parkinson, "A comparison of four computer-based telephone interviewing methods: Getting answers to sensitive questions," Behavior Research Methods, Instruments, & Computers, vol. 34, no. 3, pp. 354-363, 2002.
    [19] B. C. Choi, "Computer assisted telephone interviewing (CATI) for health surveys in public health surveillance: methodological issues and challenges ahead," Chronic Diseases and Injuries in Canada, vol. 25, no. 2, p. 21, 2004.
    [20] U. Shaw and B. Sharma, "A survey paper on voice over internet protocol (VOIP)," International Journal of Computer Applications, vol. 139, no. 2, pp. 16-22, 2016.
    [21] J. Rosenberg et al., "RFC3261: SIP: session initiation protocol," ed: RFC Editor, 2002.
    [22] M. Handley, V. Jacobson, and C. Perkins, "SDP: session description protocol," ed: RFC 2327, april, 1998.
    [23] H. Schulzrinne, S. Casner, R. Frederick, and V. Jacobson, "RFC3550: RTP: A transport protocol for real-time applications," ed: RFC Editor, 2003.
    [24] Z. Gongjian, "The study and implementation of voip intelligent voice communication system based on SIP protocol," in Proceedings of the 2016 International Conference on Intelligent Information Processing, 2016, pp. 1-9.
    [25] D. Turcato, F. Popowich, J. Toole, D. Fass, D. Nicholson, and G. Tisher, "Adapting a synonym database to specific domains," in ACL-2000 Workshop on Recent Advances in Natural Language Processing and Information Retrieval, 2000, pp. 1-11.
    [26] D. Wang, "A domain-specific question answering system based on ontology and question templates," in 2010 11th ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing, 2010: IEEE, pp. 151-156.
    [27] X. Zhang, J. Zhao, and Y. LeCun, "Character-level convolutional networks for text classification," Advances in neural information processing systems, vol. 28, pp. 649-657, 2015.
    [28] G. A. Miller, "WordNet: a lexical database for English," Communications of the ACM, vol. 38, no. 11, pp. 39-41, 1995.
    [29] W. Y. Wang and D. Yang, "That’s so annoying!!!: A lexical and frame-semantic embedding based data augmentation approach to automatic categorization of annoying behaviors using# petpeeve tweets," in Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, 2015, pp. 2557-2563.
    [30] S. Goldwater, D. Jurafsky, and C. D. Manning, "Which words are hard to recognize? Prosodic, lexical, and disfluency factors that increase speech recognition error rates," Speech Communication, vol. 52, no. 3, pp. 181-200, 2010.
    [31] M. Forsberg, "Why is speech recognition difficult," Chalmers University of Technology, 2003.
    [32] D. F. Davis and P. M. Herr, "From bye to buy: Homophones as a phonological route to priming," Journal of Consumer Research, vol. 40, no. 6, pp. 1063-1077, 2014.
    [33] 黃居仁, 謝舒凱, 洪嘉馡, 陳韻竹, 蘇依莉, 陳永祥, 黃勝偉, "中文詞彙網路:跨語言知識處理基礎架構的設計理念與實踐," 中國語文, 24卷第二期.
    [34] T. Mikolov, K. Chen, G. Corrado, and J. Dean, "Efficient estimation of word representations in vector space," arXiv preprint arXiv:1301.3781, 2013.
    [35] "PJSIP home page." [Online]. Available: https://www.pjsip.org/.
    [36] V. Këpuska and G. Bohouta, "Comparing speech recognition systems (Microsoft API, Google API and CMU Sphinx)," Int. J. Eng. Res. Appl, vol. 7, no. 03, pp. 20-24, 2017.
    [37] 注音符號-Wikipedia. Available: https://zh.wikipedia.org/wiki/%E6%B3%A8%E9%9F%B3%E7%AC%A6%E8%99%9F.
    [38] 教育部-重編國語辭典修訂本. Available: http://dict.revised.moe.edu.tw/cbdic/.
    [39] Freepik. 2021. Server free icon. Available:https://www.flaticon.com/free-icon/server_3208727?term=server&related_id=3208727
    [40] Flat Icons. 2021. Voip free icon. Available:https://www.flaticon.com/free-icon/voip_2867727?term=voip&page=1&position=5&page=1&position=5&related_id=2867727
    [41] Freepik. 2021. Modem free icon. Available:https://www.flaticon.com/free-icon/modem_1588474?term=modem&page=1&position=29&page=1&position=29&related_id=1588474
    [42] Flat Icons. 2021. Computer free icon. Available:https://www.flaticon.com/free-icon/computer_1753751?term=computer&page=1&position=49&page=1&position=49&related_id=1753751
    [43] Photo3idea_studio. 2021. Internet free icon. Available: https://www.flaticon.com/free-icon/internet_2808557?term=internet&page=2&position=61&page=2&position=61&related_id=2808557
    [44] Srip. 2021. Cloud Network free icon. Available:https://www.flaticon.com/free-icon/cloud-network_1269701?term=cloud%20network&page=1&position=28&page=1&position=28&related_id=1269701
    [45] Freepik. 2021. Route free icon. Available:https://www.flaticon.com/free-icon/route_834054?term=route&page=2&position=54&page=2&position=54&related_id=834054

    下載圖示
    2026-08-10公開
    QR CODE