簡易檢索 / 詳目顯示

回結果列表
研究生： 邱品毓
Chiu, Pin-Yu
論文名稱： 基於醫療保健知識庫之關聯路徑建立健康諮詢聊天機器人平台
A Health-Advisory Chatbot Based on Relation Path Searching of Healthcare Knowledge Base
指導教授： 盧文祥
Lu, Wen-Hsiang
學位類別： 碩士
Master
系所名稱： 電機資訊學院 - 資訊工程學系
Department of Computer Science and Information Engineering
論文出版年： 2020
畢業學年度： 108
語文別： 英文
論文頁數： 47
中文關鍵詞： 醫療保健 、知識庫 、知識圖譜 、聊天機器人
外文關鍵詞： healthcare, knowledge base, knowledge graph, chatbot
相關次數： 點閱：179  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

    • 隨著網路的快速發展與普及，使用快速、自動化且即時的聊天機器人服務成為一種趨勢。其中知識庫作為一個聊天機器人查詢知識的媒介，發展亦受到關注。
    但我們在觀察後發現，當前醫療相關之聊天機器人常因知識量不足而無法回答較為廣泛的問題。同時我們發現透過疾病與症狀的關聯來推論問題的答案，例如心臟病造成咳嗽，咳嗽不能吃冰，因而推論出心臟病不能吃冰，然而當前並沒有系統妥善利用該特性。
    為了解決上述問題，本研究提出了經由實體與詞性SDVO之結構抽取疾病、症狀及建議的方法，並建立醫療保健知識庫。我們也建構一套演算法，利用知識圖譜中的尋徑問題實作上述的問題推論，並據此建立一個聊天機器人來回答使用者醫療保健相關的問題。

    With the widespread of Internet, chatbots becomes a trend for its fast, automatic, and real-time service. At the same time, the knowledge base also become popular as a medium for chatbot to query knowledge.
    We observed that the present medical related chatbot often cannot answer wider question because of the leak of knowledge. Also, we found that we can infer the answer via the relation between disease and symptom, but there is no system utilized this feature. For example, we know that heart disease causes cough, and those who coughs cannot eat ice, thus we can infer that those who has heart disease cannot eat ice.
    To solve the problems mentioned above, we propose using SDVO pairs to extract the relation between disease, symptom, and other entities, then build a healthcare knowledge base with the extracted relations. We also designed an algorithm to implement the answer inference via the path finding of knowledge graph. After that we build a chatbot with the algorithm to answer user’s healthcare-related questions.

    證明 I 摘要 III Abstract IV 致謝 V Table of contents VI List of Tables IX List of Figures X Chapter 1 Introduction 1 1.1 Background 1 1.2 Motivation 2 1.3 Method 3 1.4 Contribution 4 1.5 Organization of this Dissertation 4 Chapter 2 Related Work 5 2.1.1 Studies on Relation Extraction 5 2.1.2 Studies on Natural Language Understanding 5 2.1.3 Studies on Natural Language Generation 6 Chapter 3 Method 7 3.1 System Architecture 7 3.2 Data Sources 8 3.2.1 Online Medical Consultation Corpus 8 3.2.2 Healthcare Lexicons 10 3.3 Corpus Preprocessing 11 3.3.1 Punctuation Correction 11 3.3.2 CKIP Natural Language Processing Tool 11 3.3.1 Correction of Word Segmentation 12 3.4 Healthcare Knowledge Graph Construction 13 3.4.1 Observing on Healthcare suggestions 13 3.4.2 The Suggestion Adverbs 15 3.4.1 The Verb of Relations 18 3.4.2 Syntactic Pattern Summarization 21 3.4.3 Healthcare Relation Extraction 21 3.5 Healthcare Knowledge base 25 3.5.1 Building Knowledge Base Using Neo4j 25 3.6 Semi-Automatic Lexicons Feedback 26 3.6.1 Lexicons Feedback 26 3.6.2 Synonym Altering 26 3.7 NLU Module 27 3.7.1 Intent Classification 27 3.7.2 Entity Recognition 27 3.8 Knowledge Matching 30 3.8.1 The Representativeness Score 30 3.8.2 Disease-Symptom-Suggestion Path Finding 31 3.8.3 Cause of Disease 35 3.8.4 Disease Similarity 36 3.9 NLG Module 37 Chapter 4 Experiments 38 4.1 Dataset 38 4.2 Evaluation Metrics 38 4.3 Evaluation on Healthcare Entity Recognition 40 4.3.1 Experiment Setup 40 4.3.2 Experiment Result 40 4.4 Evaluation on Relation Extraction 42 4.4.1 Experiment Setup 42 4.4.2 Experiment Result 42 4.5 Evaluation on Intent Classification 43 4.5.1 Experiment Setup 43 4.5.2 Experiment Result 43 Chapter 5 Conclusions and Future Work 45 5.1.1 Conclusions 45 5.1.2 Future Work 45 Chapter 6 Reference 46

    [1] "Google Trends," [Online]. Available: https://trends.google.com.tw/trends/explore?date=today%205-y&q=chatbot.
    [2] "Chatbot Market Size Worth $1.25 Billion By 2025," Grand View Research, [Online]. Available: https://www.grandviewresearch.com/press-release/global-chatbot-market.
    [3] "21 Vital Chatbot Statistics for 2020," [Online]. Available: https://outgrow.co/blog/vital-chatbot-statistics.
    [4] "CkipTagger," [Online]. Available: https://github.com/ckiplab/ckiptagger.
    [5] Y.-H. Tseng, L.-H. Lee, S.-Y. Lin, B.-S. Liao, M.-J. Liu, H.-H. Chen, O. Etzioni and A. Fader, "Chinese Open Relation Extraction for Knowledge Acquisition," 2014.
    [6] W.-C. Wu, A Health-preserving Suggestions Chatbot based on Disease Knowledge-graph, 2019.
    [7] S.-T. Chuang, Product Recommendation BOT Based On Related Task Structure Using News Event and Yahoo Answer, 2018.
    [8] T. Bocklisch, . J. Faulkner, . N. Pawlowski and A. Nichol, "Rasa: Open Source Language Understanding and Dialogue Management," 2017 .
    [9] J.-C. Cai, Summary Generation for Chinese Patient Complaint based on Medical Entity Recognition and Medical Terminology Mapping, 2019.
    [10] "快速问医生有问必答网," [Online]. Available: https://www.120ask.com/.
    [11] "5151線上照護網," [Online]. Available: http://5151.tw/.
    [12] "台灣中醫網," [Online]. Available: http://www.tcm.tw/index.php.
    [13] "食品營養成分資料庫(新版)," Taiwan Food and Drug Aministration, [Online]. Available: https://consumer.fda.gov.tw/Food/TFND.aspx?nodeID=178.
    [14] "體育運動列表," wikipedia, [Online]. Available: https://zh.wikipedia.org/wiki/體育運動列表.
    [15] "穴位," A+醫學百科, [Online]. Available: http://cht.a-hospital.com/w/穴位.
    [16] "E-HowNet," [Online]. Available: http://ehownet.iis.sinica.edu.tw/.
    [17] "Neo4j," Neo4j, Inc., [Online]. Available: https://neo4j.com/.
    [18] "Cypher Query Language," Neo4j, Inc., [Online]. Available: https://neo4j.com/developer/cypher-query-language/.
    [19] "Chatbot Report 2019: Global Trends and Analysis," [Online]. Available: https://chatbotsmagazine.com/chatbot-report-2019-global-trends-and-analysis-a487afec05b.
    [20] S. Patel, "Top 12 Chatbots Trends and Statistics to Follow in 2020," [Online]. Available: https://www.revechat.com/blog/chatbots-trends-stats/.

    無法下載圖示 校內：2025-08-14公開
    校外：不公開
    電子論文尚未授權公開，紙本請查館藏目錄
    QR CODE