- 期刊
AI聊天機器人在澎湖旅遊諮詢之開發與應用
Development of AI Chatbot and Its Application in Penghu Tourism Consulting
摘要
人工智慧（AI）被視為下一代IT系統的核心，未來的市場相當龐大。AI新經濟正夯，全球各大科技廠商紛紛推出各具特色的聊天機器人，AI將會是服務業重要的幫手，藉由AI聊天機器人可隨時與顧客進行及時的交流與回覆。本研究透過AI及語料庫，發展了一套專注於澎湖當地觀光休閒產業所用的聊天機器人系統，首先使用Dialogflow框架構建AI應用程式，接續將其導入至LINE社群通訊軟體來完成互動式介面。此應用投入市場予以運作、測試、評估及改善。透過331位遊客的實際使用、測試與學習，第一版之資料庫的正確率為66.9%，經強化Intent（意圖）之設計及修正後，第二版的正確率提升至77.7%及第三版的87.3%，歷經四次的調整之後，應用程式的正確率整體正確率超過80%。本研究同時設計問卷來蒐集及分析使用者對應用程式的使用心得及旅遊偏好，由後台之資料可統計得觀光客之搜尋的關鍵熱點，該等資訊再與遊客背景資料進行比對分析。研究結果顯示，詢問度最高之項目為「澎湖餐廳」，年齡層為20-29歲。本應用程式可結合當地政府機關或是相關產業，進行觀光政策及活動推播之行銷，亦可協助商家瞭解客戶之意圖及需求，進而提高客服滿意度與業務效率。透過實証成果的資料整理，更有助於強化自然語言理解（Natural Language Understanding, NLU）的Intent（意圖）設計，因而能使機器學習（Machine Learning, ML）的準確率再行提升。
並列摘要
Artificial intelligence (AI) is regarded as the core of the next generation of IT system, and the market in the future is quite huge. With the new economy of AI, many major technology manufacturers around the world have launched their own distinctive chat robots. AI will be an important helper in the service industry. With AI chatbots, customers can communicate and reply in time at any time. Through AI and text corpus, this study developed a chatbot system focused on the local sightseeing and leisure industry in Penghu, first using the Dialogflow framework to build a AI application, and then importing it into LINE community communication software to complete the interactive interface. This application is put into the market for operation, testing, evaluation and improvement. Through the actual use, testing and learning of 331 tourists, the accuracy rate of the first version of the database was 66.9%. After the design and correction of the enhanced Intent, the accuracy rate of the second version was increased to 77.7% and the third 87.3% of the version, after four adjustments, the overall accuracy rate of the application is over 80%. In this study, a questionnaire was designed to collect and analyze the user's experience and travel preference of the application, and the key hot spots of the tourists' search could be obtained from the background data. The results show that the most inquired item is Penghu restaurant, aged 20-29 years old. This application can promote tourism policies and activities in combination with local government agencies or related industries. It can also help businesses understand customers' intentions and needs, so as to improve customer service satisfaction and business efficiency Through the data collation of empirical results, it is more helpful to strengthen the intent design of natural language understanding (NLU), so as to improve the accuracy of machine learning (ML).