- 學位論文
基於自然語言轉換SQL之資料庫語音查詢機器人
Translating Voice Message to SQL based on Nature Language Processing
摘要
隨著資料庫的資料日益增加，資料的複雜度也逐漸提升，以過往的查詢方式，過去的查詢是手動的點選網頁方式，需要倚賴使用者對問題的了解程度，而且需要自行篩選條件，才能得知答案，不但浪費時間，也增加操作的複雜度。 因此，利用人工智慧的技術，提供一個直覺化和方便性的查詢方式，讓使用者省時方便得到想要的查詢結果，便成為很重要的研究議題。 本論文採用自然語言處理(Natural Language Processing, NLP)的技術，利用深度神經網路建立模型，將自然語言轉換成對應的資料庫SQL語法。讓使用者輸入時，只要以語音來隨意輸入詢問的語句，我們深度學習的模型便能將語音轉成文字，再將文字轉換成資料庫相對應的「欄位」、「關係」與「值」的查詢欄位，最後轉換成SQL的語法，查詢到使用者需要的正確結果。
並列摘要
In the past query methods, the user's understanding of the question needs to be relied on, and the user needs to filter the conditions to get the answer. In the past, queries were manually clicked on web pages, which not only wasted time, but also increased the complexity of operations. Therefore, the use of artificial intelligence technology to provide an intuitive and convenient query method so that users can save time and easily obtain the desired query results has become a very important research topic. This thesis adopts Natural Language Processing (NLP) technology and uses deep neural networks to build models, and convert natural language into corresponding database SQL syntax. When allowing users to use, as long as they use voice to say the query sentence at will, our deep learning model can convert the voice into text, and then transfer the text into the query fields of the corresponding "field", "relation" and "value" in the database, and finally convert it into SQL Therefore, the query results can be directed presented for the user’s query.