臺灣博碩士論文加值系統

近年來即時通訊平台相繼推出聊天機器人API，使聊天機器人(Chatbot)與自然語言處理技術成為人們熱門議題之一，各企業相繼推出專屬的聊天機器人，這些聊天機器人往往缺乏智能。基金數量逐年增加，讓投資人在抉擇上出現困擾，如何建立一個具備智能的聊天機器人來找出適合投資人投資的基金是本研究所要探討的議題。本研究提出一個整合聊天機器人與人工智慧於雲端運算架構(Integrating Chatbot and Artificial Intelligence into Cloud Computing Framework, ICAICCF)來解決上述問題。將機器學習中支援向量回歸(Support Vector Regression, 簡稱SVR)以及語意網(Semantic Web)架設於雲端平台作為基本架構，並透過人工智能標籤語言(Artificial Intelligence Markup Language, 簡稱AIML)技術來加強聊天機器人自然語言能力。本研究開發出基金智能雲端平台(Fund Intelligent Cloud Platform, FICP)來驗證ICAICCF的可行性。首先透過Line建置聊天機器人，並透過Open Data所提供的資料製作Line bot基金問卷，取得投資人的投資個性以及基金類型喜好，再經由Semantic Web找出適合投資人投資的基金，最後根據SVR演算法預測下期基金漲跌幅度將基金排序推薦投資人前三名的基金，運用AIML技術讓使用者透過對話方式取得基金問題的答案。為了加快運算效率，本研究透過Spark、Storm雲端運算將語意網、機器學習架設於雲端上，並比較兩者之間的效率，根據實驗結果顯示Storm適用於小資料塊即時運算，Spark適合大資料量的批量運算。根據使用者在滿分為5分的系統評估問卷調查結果顯示，使用者平均滿意度為4.173分，證實ICAICCF的可行性。

In recent years, real-time communication software such as Facebook and Line have launched Chatbot API sequentially, which makes Chatbot and natural language processing technology become one of the hot topics. Each enterprise has launched exclusive Chatbot one after another, which is usually short of intelligence. The number of funds is increasing year by year, which makes investors have trouble in choosing. Building a Chatbot with intelligence to find the fund suitable for investors is the topic of this study. The study proposes integrating Chabot and Artificial Intelligence into Cloud Computing Framework (ICAICCF) to solve the above problem. Support Vector Regression (SVR) and Semantic Web (Semantic Web) in Machine Learning are erected on cloud platform as the basic framework, and Artificial Intelligence Markup Language (AIML) is used to strengthen the natural language ability of Chatbot. The study develops Fund Intelligent Cloud Platform (FICP) to verify the feasibility of ICAICCF. Firstly, build Chatbot through Line, develop Line bot fund recommendation through the information provided by Open Data, and acquire the investor's investment personality and fund type preference. Secondly, find out the fund suitable for the user's investment through Semantic Web. Finally, predict the next fund fluctuation according to the SVR algorithm, recommend the top three funds to the user, and use AIML technology to let the user acquire the answer of fund question through the conversational mode. In order to speed up the computing efficiency, the study erects Semantic Web and Machine Learning on the cloud through Spark and Storm cloud computing, and compares the efficiencies of the two. According to the experimental results, Storm is suitable for small data block real-time operation, and Spark is suitable for batch operation of large data amount. According to the results of the system evaluation questionnaire with a full score of 5, the users' average satisfaction is 4.173, which verifies the feasibility of ICAICCF.

摘要......................i
Abstract......................ii
誌謝......................iii
目錄......................iv
表目錄......................vi
圖目錄......................vii
第一章 簡介......................1
1.1 研究背景......................1
1.2 研究動機......................2
1.3 研究目的......................2
第二章 研究技術探討......................3
2.1 聊天機器人......................3
2.2 Semantic Web......................4
2.3.1 RDF......................5
2.3.2 Ontology......................5
2.3.3 Rule......................5
2.3.4 Inference Engines......................5
2.3 支援向量回歸(Support Vector Regression)......................6
2.4 AIML人工智能標籤語言......................9
2.5 Hadoop......................11
2.5.1 Hadoop HDFS......................12
2.5.2 Hadoop YARN......................13
2.6 Spark......................15
2.6.1 Spark RDD......................16
2.6.2 Spark叢集管理器......................17
2.7 Storm......................18
2.7.1 Topology......................19
2.7.2 Storm叢集管理器......................20
2.8 相關研究......................21
第三章 系統設計與研究步驟......................22
2.1 系統架構......................22
3.1.1 一般性架構......................22
3.1.2 基金智能雲端平台......................24
2.2 研究方法與流程......................26
3.2.1 Line聊天機器人......................26
3.2.2 AIML人工智能標籤語言......................29
3.2.3 整合AIML以及聊天機器人......................33
3.2.4 語意網......................35
3.2.5 資料預處理......................42
3.2.6 支援向量回歸......................45
3.2.7 整合系統......................48
第四章 系統實作......................49
4.1 應用場景......................49
4.2 系統特色......................49
4.3 Line bot 使用畫面......................50
第五章 系統實驗與評估......................53
5.1 雲端運算環境架構......................53
5.1.1 測試環境......................53
5.1.2 雲端運算叢集模式......................56
5.2 雲端運算效能測試......................58
5.2.1 測試資料......................58
5.2.2 測試比較......................59
5.3 測試總結......................62
5.4 系統評估......................64
第六章 結論......................65
參考文獻......................66
Extended Abstract......................69

[1]“【證券經紀商投資人開戶統計表】年報 - TWSE 臺灣證券交易所.” [Online]. Available: http://www.twse.com.tw/zh/statistics/statisticsList?type=07&subType=261. [Accessed: 28-Jan-2019].
[2]N.Albayrak, A.Özdemir, andE.Zeydan, “Yapay Zeka Tabanlı Rehber Robotlara Genel Bir Bakı ¸ s ve Örnek Bir Rehber Robot Uygulaması An Overview of Artificial Intelligence Based Chatbots and An Example Chatbot Application,” 2018 26th Signal Process. Commun. Appl. Conf., pp. 1–4.
[3]O.Vinyals, T.Koo, andG.Hinton, “Grammar as a Foreign Language,” pp. 1–9.
[4]“弱人工智能 - Wikipedia.” [Online]. Available: https://en.wikipedia.org/wiki/Weak_AI. [Accessed: 28-Jan-2019].
[5]R.VYampolskiy, “Turing Test as a Defining Feature of AI-Completeness,” pp. 3–17, 2012.
[6]S. A.Abdul-Kader andJ.Woods, “Survey on Chatbot Design Techniques in Speech Conversation Systems,” IJACSA) Int. J. Adv. Comput. Sci. Appl., vol. 6, no. 7, 2015.
[7]A.Augello, G.Saccone, S.Gaglio, andG.Pilato, “Humorist bot: Bringing computational humour in a chat-bot system,” Proc. - CISIS 2008 2nd Int. Conf. Complex, Intell. Softw. Intensive Syst., pp. 703–708, 2008.
[8]V.Mala, “Semantic and Keyword Based Web Techniques in Information Retrieval,” 2016 Int. Conf. Comput. Commun. Autom., pp. 23–26, 2016.
[9]L.Gai, X.Wang, andT.Wang, “Leveraging Neighborhood Summaries for Efficient Queries on Existing Relational RDF Systems,” 2017.
[10]X.Yuan, “Combining Ontologies for Requirements Elicitation,” 2015 IEEE Int. Model. Requir. Eng. Work., pp. 1–5, 2015.
[11]Chang Choi, Miyoung Cho, Eui-young Kang, andPankoo Kim, “Travel Ontology for Recommendation System based on Semantic Web,” 2006 8th Int. Conf. Adv. Commun. Technol., pp. 624–627, 2006.
[12]W.-K.Tsai, “碩士論文 專家系統在鄰里遊憩公園 設施評估應用之研究,” 臺灣博碩士論文, p. pp.1-96, 2010.
[13]R.Harimurti, Y.Yamasari, andI. G. P. B.Asto, “Predicting Student ’ s Psychomotor Domain on The Vocational Senior High School Using Linear Regression,” 2018 Int. Conf. Inf. Commun. Technol., pp. 448–453, 2018.
[14]J.Liu andG.Geng, “Fault Prediction for Power Plant Equipment Based on Support Vector Regression,” 2015 8th Int. Symp. Comput. Intell. Des., vol. 2, pp. 461–464, 2015.
[15]W.Yi, “An Intelligent Algorithm of Support Vector Regression Parameters Optimization in Soft Measurements,” 2016 XIX IEEE Int. Conf. Soft Comput. Meas., no. 1, pp. 404–406.
[16]F. A.Mikic, J. C.Burguillo, M.Llamas, D. A.Rodríguez, andE.Rodríguez, “CHARLIE: An AIML-based chatterbot which works as an interface among INES and humans,” 20th EAEEIE Annu. Conf. EAEEIE 2009 - Form. Proc., 2009.
[17]B. R.Ranoliya, N.Raghuwanshi, andS.Singh, “Chatbot for university related FAQs,” 2017 Int. Conf. Adv. Comput. Commun. Informatics, ICACCI 2017, vol. 2017–Janua, pp. 1525–1530, 2017.
[18]P.Merla, “Data Analysis using Hadoop MapReduce Environment,” pp. 4783–4785, 2017.
[19]N.Alasmari, “Optimising Cloud-Based Hadoop 2 . x Applications,” 2018 IEEE/ACM Int. Conf. Util. Cloud Comput. Companion (UCC Companion), pp. 43–46, 2018.
[20]S.Singh, “Analysis of HDFS RPC and Hadoop with RDMA by Evaluating Write Performance,” 2016 6th Int. Conf. - Cloud Syst. Big Data Eng., pp. 368–372, 2016.
[21]J. S.Manjaly, “A comparison Study and performance evaluation of Schedulers in Hadoop YARN,” no. Icces, pp. 78–83, 2017.
[22]J.Ps, S.Business, S.Pvt, andP.Samuel, “Analysis and modeling of Resource Management Overhead in Hadoop YARN Clusters,” 2017.
[23]D.Niu, B.Chen, andZ.Chen, “The Classified and Active Caching Strategy for Iterative Application In Spark,” 2018 27th Int. Conf. Comput. Commun. Networks, pp. 1–2, 2018.
[24]A.Kuzmanovska, H.Van DenBogert, R.Mak, andD.Epema, “Achieving Performance Balance among Spark Frameworks with Two-Level Schedulers,” 2018 18th IEEE/ACM Int. Symp. Clust. Cloud Grid Comput., pp. 133–142, 2018.
[25]M.Jayaratne andD.Alahakoon, “Apache Spark based Distributed Self-Organizing Map Algorithm for Sensor Data Analysis.”
[26]B.Zhou, “Machine-Learning-Based Online Distributed Denial-of-Service Attack Detection Using Spark Streaming,” 2018.
[27]R. B.Evans, “Apache Storm a Hands on Tutorial,” 2015 IEEE Int. Conf. Cloud Eng., p. 2, 2015.
[28]D.Im, C.Cho, andI.Jung, “Detecting a Large Number of Objects in Real-time Using Apache Storm,” 2014 Int. Conf. Inf. Commun. Technol. Converg., pp. 836–838, 2014.
[29]S.Son, S.Lee, M.Gil, M.Choi, andY.Moon, “Locality Aware Traffic Distribution in Apache Storm for Energy Analytics Platform,” 2018 IEEE Int. Conf. Big Data Smart Comput., pp. 721–724, 2018.
[30]Y.Jeong, S.Kim, andB.Yoon, “An Algorithm for Supporting Decision Making in Stock Investment through Opinion Mining and Machine Learning,” 2018 Portl. Int. Conf. Manag. Eng. Technol., pp. 1–10, 2018.
[31]“2018 2nd International Conference on Data Science and Business Analytics,” pp. 1–6, 2018.
[32]Y.Xia, Y.Liu, andZ.Chen, “Support Vector Regression for Prediction of Stock Trend,” 2013 6th Int. Conf. Inf. Manag. Innov. Manag. Ind. Eng., vol. 2, pp. 123–126, 2013.
[33]T. B.Pun andT. B.Shahi, “Nepal Stock Exchange Prediction Using Support Vector Regression and Neural Networks,” 2018 Second Int. Conf. Adv. Electron. Comput. Commun., pp. 1–6, 2018.
[34]G.Saqib, “Intelligent Chatting Service Using AIML,” 2018 Int. Conf. Curr. Trends Towar. Converging Technol., pp. 1–3, 2018.
[35]P. A.Ginige, “based sessions for diabetes education,” 2018 32nd Int. Conf. Adv. Inf. Netw. Appl. Work., pp. 698–703, 2018.
[36]M.Wen, “A Conversational User Interface for Supporting Individual and Group Decision- Making in Stock Investment Activities Ming-Hui Wen,” 2018 IEEE Int. Conf. Appl. Syst. Invent., pp. 216–219, 2018.
[37]S.Arsovski, B.Markoski, P.Pecev, L.Ratgeber, andN.Petrov, “An ontology driven credit risk scoring model,” 2014 IEEE 15th Int. Symp. Comput. Intell. Informatics, pp. 301–305, 2014.
[38]M.Tirea, “Text Mining News System - Quantifying Certain Phenomena Effect on the Stock Market Behavior,” 2015 17th Int. Symp. Symb. Numer. Algorithms Sci. Comput., pp. 391–398.
[39]I.Hadoop, M. A.Hedjazi, I.Kourbane, Y.Genc, andB.Ali, “Büyük Ölçekli Resim Sınıflandırması ˙ Spark ve Storm Kar ¸ stırılması A Comparison of Hadoop , Spark and Storm for the Task of Large Scale Image Classification,” 2018 26th Signal Process. Commun. Appl. Conf., pp. 1–4.
[40]W.Sukfong, S.Nakkrasae, andP.Panpipat, “Motor insurance knowledge management in AIML format using tree structured dictionary mechanism and conceptual graph,” 2016 8th Int. Conf. Knowl. Smart Technol. KST 2016, pp. 158–163, 2016.