Archives

期刊全文

Implementation and Evaluation of a Retrieval-based Chinese Humor Chatbot

基於檢索方法的中文幽默對話系統之建置應用與評估

Volume / Date

 VOL. 18 (NO. 2) / DEC. 2020

Page

 pp. 73-101

DOI

 10.6182/jlis.202012_18(2).073

Language

 Chinese (with English extended abstract)

Yuen-Hsien Tseng 曾元顯*、Wei-Lun Hsu 許瑋倫、Wun-Syuan Wu 吳玟萱、Yi-Ciao Gu 古怡巧、Hsueh-Chih Chen 陳學志
* 通訊作者

Yuen-Hsien Tseng 曾元顯* Graduate Institute of Library & Information Studies, National Taiwan Normal University, Taipei, Taiwan
國立臺灣師範大學圖書資訊學研究所
MOST AI Biomedical Research Center, Tainan, Taiwan
科技部人工智慧生技醫療創新研究中心
ORCID: 0000-0001-8904-7902
samtseng@ntnu.edu.tw
Wei-Lun Hsu 許瑋倫 Graduate Institute of Library & Information Studies, National Taiwan Normal University, Taipei, Taiwan
國立臺灣師範大學圖書資訊學研究所
Wun-Syuan Wu 吳玟萱 Graduate Institute of Library & Information Studies, National Taiwan Normal University, Taipei, Taiwan
國立臺灣師範大學圖書資訊學研究所
Yi-Ciao Gu 古怡巧 Graduate Institute of Library & Information Studies, National Taiwan Normal University, Taipei, Taiwan
國立臺灣師範大學圖書資訊學研究所
Hsueh-Chih Chen 陳學志 Department of Educational Psychology and Counseling, National Taiwan Normal University, Taipei, Taiwan
國立臺灣師範大學教育心理與輔導學系
MOST AI Biomedical Research Center, Tainan, Taiwan
科技部人工智慧生技醫療創新研究中心
ORCID: 0000-0003-2043-0190

Abstract

This research is to construct a humorous corpus, develop related technologies, implement a retrieval-based “icebreaker chatbot” system which allows users to find relevant jokes for use in relaxing an unduly formal atmosphere when interacting with people, and finally evaluate its effectiveness. Through the iterative steps of the information system development research method, query expansion based on Word2Vec technology, frequent keyword prompts, and random recommendation of good jokes are added after user feedback. The results are that the proportion of user queries that fail to find jokes is reduced from 25.4% to 16.7% and that the icebreaking effect achieved has been increased from 27.9% to 39.9%. The importance of this research not only compiled a corpus of nearly 5,000 Chinese jokes, but also built a Chinese humor dialogue system, which have both been publicized at https://github.com/SamTseng/icebreaker for future use and verification. Empirical experience and implications of this study include: automating the richness and quality of joke corpus and providing recommendation service are important R & D efforts to improve the effectiveness of such services.


本研究發展幽默語料庫，開發相關的技術，實作一個基於檢索方法的「破冰機器人」系統，讓使用者透過對話找出相關的笑話，在與人互動時打破冰冷的氣氛，活絡陌生、尷尬的情境，最後評估其運用成效。透過資訊系統開發研究法的循環步驟，經過回饋後加入 Word2Vec的查詢擴展、關鍵詞查詢提示，以及好笑話的隨機推薦等功能，讓使用者找不 到笑話的比例從25.4%降低到16.7%，而系統達到的破冰效果從27.9%提升到39.9%。綜合而言，本研究不僅蒐集編製了近5,000則正體中文幽默語料庫，也建置中文幽默對話應用系統，語料與程式公開於https://github.com/SamTseng/icebreaker。本研究結論提供了實證經驗與意涵:自動化豐富笑話語料並確保其幽默程度，以及提供推薦功能，是提升此類服務成效的重要研發工作。

Keywords

Computational Humor; Chinese Humorous Dialogue; Humor Corpus; Dialogue System; Icebreaker Chatbot

計算幽默、中文幽默對話、幽默語料、對話系統、破冰機器人